Development of an observer-reported outcome measure to capture the signs and impact of fever distress symptoms in infants and young children

This qualitative study aimed to construct an observer-reported outcome measure (ObsRO) to evaluate fever distress in young children. A literature review was conducted to identify fever-related concepts. Clinical experts were interviewed for feedback on these concepts. Parents of young children were interviewed to identify behaviours the child exhibited during a recent fever episode. Fever sign and behaviour concepts endorsed by ≥ 20% parents were used to create items for the draft ObsRO. Parents of young children who recently had fever completed the ObsRO and gave feedback during two successive rounds of cognitive interviews. Twenty-five parents participated in the concept elicitation. Mean child age was 2.7 years (range: 0.6–5.8 years). Fever sign and behaviour concepts endorsed by ≥ 20% participants were high temperature (80%), skin hot to touch (32%), skin redness/flushing (32%), reduced appetite/drink (96%), needy/clingy/irritable (48–92%), less active/interactive (68–84%) and lethargic (64–88%). Eighteen items, four in the Fever Signs Module and 14 in the Fever Behaviours Module, were developed for the draft ObsRO. Chosen recall period was 24 h. Thirty participants (Round 1: n = 17; Round 2: n = 13), participated in cognitive interviews. Mean child age was 2.4 years (range 0.3–5.8). Round 1 feedback resulted in two Fever Signs items being combined. Three Fever Behaviour items were deleted, six revised and four unchanged. No changes were made following Round 2 feedback. Most participants understood all aspects of the ObsRO and found it user-friendly. The ObsRO will undergo further development in validation studies testing measurement properties of each item.


Introduction
Pre-school children suffer frequent episodes of illness leading to more primary care consultations than any other age group. The most common reasons are coughs, colds, earache and fevers [1] with fever as the primary presentation for these illnesses. Low-grade fever is common in young children and is usually a symptom of infection. The underlying cause of childhood fever is generally benign, and fever has a beneficial effect in terms of fighting infection [2]. However, fever can cause distress and discomfort in children, leading to a high degree of parental concern; parents may believe that the fever itself is a disease or a harmful symptom in its own right [2,3]. This common, albeit understandable, misconception (sometimes referred to as 'fever phobia') may lead to parental uncertainty and anxiety in managing fever in their children [2,[4][5][6]. As a result, fever in a child commonly leads to unscheduled physician visits, telephone calls by parents to healthcare professionals for advice on fever control. Prospective studies in the United Kingdom (UK) have demonstrated that around 72% of pre-school children attend their family practitioner within the first 5 years with fever being at least of the primary reasons for attending in 25% of all cases [1]. Furthermore, estimates suggest that between 10 and 19% of out-of-hour family practitioner visits and 14% of emergency room visits in the UK for infants and children aged > 3 months are for fever [7][8][9]. Nevertheless, healthcare bodies do recognise the importance of parental concerns as part of the management of paediatric fever. The UK National Institute for Health and Care Excellence (NICE) states that 'reported parental perception of a fever should be considered valid and taken seriously by healthcare professionals', as even mild symptoms may indicate a life-threatening infection [9].
For febrile children without any indication of a serious underlying condition ('low-risk' fever), and key among the updated recommendations, is the need to treat symptoms of the fever with a focus on comforting the child, rather than on achieving normothermia. National guidelines exist for the management of fever in infants [10] and recommend home management [2,[10][11][12] and childhood fever associated with discomfort or pain can be easily treated with over the counter (OTC) anti-pyretic (and analgesic) agents. For instance, NICE guidelines for management of fever in children aged < 5 years recommend using acetaminophen/paracetamol or ibuprofen in children with fever only if the child appears distressed [9] with the aim of alleviating distress rather than reducing body temperature. The concept of 'distress' is difficult to define holistically as paediatric verbal articulation of distress and associated behaviours may vary across different age groups and may also depend on the caregiver's perception. Similar guidance in making reference to distress or discomfort is seen across many other countries [11].
In its guidelines, NICE recognises that there is a 'need to improve the recognition, assessment and immediate treatment of feverish illnesses in children' [9]. However, self-reported information from young children may not be reliable across all age groups and is virtually impossible to obtain from infants. Furthermore, there are currently no validated observer-reported outcome measures (ObsRO) available to evaluate fever distress in young children. We therefore describe a qualitative, cross-sectional study to develop an ObsRO measure that evaluates fever distress in children aged ≥ 3 months to < 6 years. The ultimate aim of this ObsRO is to enable parents and caregivers to measure fever distress in their child to help their decision-making process in determining the frequency of treatment with over-the-counter medicines as well as facilitate interactions with healthcare practitioners, such as family doctors and community pharmacists.

Study design
This study comprised four stages (Fig. 1). In Stages 1 and 2 (concept identification and elicitation, respectively), a conceptual framework for fever distress in children was developed based on scientific literature and feedback from both clinical experts and parents/caregivers of young children with fever. The most common signs and behaviours of fever distress were used to develop concept items for the draft ObsRO instrument (Stage 3). Parents of young children with fever were recruited for participation in cognitive interviews whereby they completed an electronic version of the draft instrument and gave feedback on the usability of the instrument and relevance of the items (Stage 4).
All interviews were conducted on behalf of the sponsor (Reckitt Benckiser, RB) by a third-party agency (Evidera) in the UK by staff trained in qualitative interview techniques. The study protocol and informed consent forms (ICs) were approved by the UK Health Research Authority (NHS HRA South West -Central Bristol Research Ethics Committee, reference 249111). All documentation prepared as part of this work was carried out in accordance with the United States (US) Food and Drug Administration (FDA) patient-reported outcome (PRO) guidance recommendations and standards [13]. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Stage 1: concept identification
Targeted research studies, clinical articles and guidance documents were reviewed to identify fever-related concepts and create a definition of 'fever distress'. Signs of fever were conceptualised as those confirming the presence of fever (e.g., red cheeks). Fever distress was conceptualised as fever-related changes in behaviour (e.g., loss of appetite). The initial conceptual framework for fever distress included behaviours (restlessness/agitation, apathy/limpness and loss of appetite) and signs (red cheeks and glazed eyes). A conceptual disease model (CDM) was developed to understand fever distress. The authors interviewed three expert clinicians, two from the UK and one from the US, for feedback regarding the above concepts and the CDM. These clinical experts comprised two consultant paediatricians (UK and US) and one general practitioner; all clinical participants had 10 or more years of clinical experience in the management and treatment of paediatric fever.

Recruitment
Participants were screened for eligibility using ethics committee-approved screening forms. Eligible participants were parents/caregivers of children aged ≥ 3 months to < 6 years who had recently resolved fever (within the previous 2 weeks). Potential participants were recruited through a third-party agency (Global Perspectives). Parents/caregivers were recruited purposively (i.e. non-random sampling) via this agency which specialised in patient recruitment for nonclinical research studies. Eligible participants were screened using an ethics-approved screening form. Participants had a good command of English, gave informed consent and were willing to be audio-recorded during the interview.
Children of the participants were grouped according to age: (1) ≥ 3 months to 1 year; (2) 1 to < 2 years; (3) 2 to < 4 years; (4) 4 to < 6 years. Telephone interviews were scheduled with eligible participants and an interview packet containing an introductory letter, two copies of the IC and a sociodemographic form was delivered to the participant by courier. Participants were reimbursed for their time.

Interview
All participants gave confirmation of written informed consent before the elicitation interview began. Each interview (including time for consent, interview and completion of case report forms) lasted approximately 60 min and was audio-recorded. All interviews were conducted by a thirdparty agency in the UK (Evidera/PPD) by staff trained in qualitative interview techniques, who had been trained specifically on the interview guide and who extensive experience in conducting qualitative interviews. The interviewer discussed parental understanding of fever, recent experience with fever, fever-related signs and behaviour, caregiver fever management, caregiver actions and impact. Participants completed the sociodemographic questionnaire after the interview and returned this, along with the two signed copies of ICs, to study personnel. A fully-signed IC was returned to participants for their personal records.

Item generation
Fever signs and behaviours endorsed by ≥ 20% of participants elicited during interviews were selected for item development. These were further evaluated against item writing criteria adapted from the International Society for Pharmacoeconomic and Outcomes Research-Patient Reported

Draft ObsRO instrument
Development of the ObsRO instrument, including creation of questions, recall period, and response choices, was guided by best practice guidance outlined in the ISPOR PRO Good Practice Task Force, Part I and Part II [14,15]. The language of participants during elicitation interviews (i.e., their descriptions of fever signs and behaviours) guided the wording of the instructions for use, item stems and response options. Potential recall periods and response options were evaluated for best fit across sign and behaviour items. The recall period was determined by the time interval between the child exhibiting the sign/behaviour and it being described by parents and clinical experts (i.e., several days, hours, etc.) and the intended use of the measure in future trials where daily fever assessment will be required.

Stage 4: cognitive interview
Cognitive interviews were conducted to test the content validity of the draft ObsRO. Parents of children with recently resolved fever were recruited via a third-party agency (Global Perspectives) using the same process as the elicitation interviews. Eligible participants were sent an IC, sociodemographic form, the ObsRO and an importance questionnaire electronically at the time of the interview.
Cognitive interviews were conducted in person, via webcam or by telephone; all sessions were audio-recorded. Interviews were performed in two rounds; in Round 1, participants completed Version 1 of the draft ObsRO instrument whereas in Round 2, participants completed Version 2. The interviewer sought feedback from participants on their understanding and comprehension of the items in the ObsRO, the timeframe and response options. Participants were also asked about usability. (The cognitive interview template is provided in Supplementary Material 2.)

Qualitative data analysis
Interview audio files were transcribed and transcripts were cleaned, de-identified and evaluated using ATLAS.ti qualitative software version 8.0 [16].
For elicitation data, a coding framework was developed to organize the text data for the major discussion topics and concepts that emerged during discussions. For cognitive interviews, feedback on instruction, response options and item meaning were coded, as well as emerging issues such as missing or confusing content. Team members were trained to use a codebook that coded words and phrases from interview participants. Two and three team members independently coded elicitation and cognitive interview transcripts, respectively, after which all codes were compared, discussed, and reconciled wherever differences occurred. A random sample of approximately 20% of coded transcripts was reviewed. If problems were identified, additional transcripts or codes were reviewed. Key concepts that emerged from coding of elicitation data were put into a concept frequency table that showed how many participants endorsed a particular concept. Concepts were also plotted on a saturation grid. Concept saturation was reached at a point when no new concepts continued to emerge.
An item tracking matrix was developed to capture participants' comprehension and relevance of each item on the ObsRO. The matrix also tracked changes made to each item during ObsRO development, providing supportive evidence and rationale for each change.

Quantitative data analysis
Tables summarizing quantitative data (means, standard deviations, and range) were presented for continuous variables and frequency and percentage for categorical variables. Quantitative information was used to characterize the study sample and document representation of the target population in accordance with regulatory guidance. In addition, descriptive statistics were calculated for the ObsRO scores.

Stage 1: concept identification
Clinicians gave feedback on the initial concept framework of fever distress. They stated that symptoms of fever can be complex, as the underlying illness causes diverse symptoms, while the body raises its temperature to combat the illness. They also stated that some signs/symptoms would present differently in children of different ages. For example, infants might have trouble feeding, swallowing and breathing, whereas children aged 4-5 years might have a sore throat, be less active, and drink and eat less. Clinicians also noted that parents would be unlikely to report fever signs such as 'depressed fontanelle' (a sign of dehydration) as this would be identified by a clinician during a physical examination.
Clinicians noted that the CDM should focus more on fever instead of immune function and physiological health factors. The preliminary fever distress CDM is presented in Fig. 2.

Interview results
Participants were asked if they understood the term 'fever'. All participants (n = 25, 100%) stated that a fever is a high temperature (higher than considered medically normal). Participants were also asked what they believe causes fever. The most common response (n = 11, 44%) was that fever is caused by an infection, whether viral or bacterial. Ten (40%) participants considered fever to be the result of an immune response against disease; body temperature rises because the body is fighting off infection.
The most common signs of fever identified by participants were objective assessment of body/skin temperature using a thermometer (n = 20, 80%), subjective assessment of body/skin temperature by touching (n = 8, 32%) and redness/flushing (n = 8, 32%; Table 2). Saturation of fever signs in the overall sample was reached by the third participant. Overall, the most common fever-associated behaviours identified by participants were reduced appetite (n = 24, 96%), needy/clingy behaviour (n = 23, 92%) and crying (n = 22, 88%; Table 2). The duration of fever ranged from 0.5 to 5 days (median 4 days). Fever signs and behaviour concepts endorsed by ≥ 20% participants are shown in Fig. 3.
The most common actions participants took when realising their child was unwell was to give medication (n = 24, 96%), offer fluids (n = 23, 92%) and food (n = 17, 68%). The most common medications participants gave their children were Calpol® (n = 19, 76%) and Nurofen® (n = 7, 28%). The majority of participants (n = 14, 56%) did not take their children to a healthcare provider because the fever symptoms were not severe. Those participants that did take their children to a healthcare provider (n = 11, 44%) were not Fig. 2 Preliminary fever distress conceptual disease model prescribed any additional medication, but were advised to treat with Calpol® or Nurofen for Children® and to monitor the child.

Stage 3: instrument development
Fever signs and behaviours endorsed by ≥ 20 participants during the elicitation interviews were further mapped to item writing criteria. The following instructions were provided to parents when completing the draft instrument: The following questions assess how fever has affected your child in the past 24 h. Please answer all questions by selecting the one option that best describes your observations of your child.
A 24-h recall period was selected for both signs and behaviours items. A modular approach was developed for the draft questionnaire; participants would first answer questions about fever signs (Fever Signs Module) to ascertain whether or not the child had fever before answering questions about the child's behaviour (Fever Behaviour Module). The Fever Signs Module contained four items which were simple yes/ no questions to indicate presence/absence of a sign in the previous 24 h. The Fever Behaviours Module contained 14 items which allowed participants to compare their child's behaviour in the past 24 h to how they behaved normally. Four responses were possible for each item: more than usual; the usual; less than usual; not at all.
The draft ObsRO instrument was adapted to a web-based platform for participants to complete on any electronic device.

Interview results
Round 1 feedback All participants understood the instructions and thought they were clear; none gave any suggestions for improvement. Eight (47%) participants felt that the 24-h recall period was appropriate. However, when taking into account the duration of fever and administration of medicine, participants felt the recall period could be lengthened. No modifications were made to the recall period or response options after Round 1 of the cognitive interviews. Two (12%) participants felt that response options were inappropriate for some of the items (e.g., it is unlikely that a child with fever will play 'more than usual') and one suggested changing responses to yes/no for some items.
Most participants understood all four Fever Signs items and felt they were important. However, nine (53%) participants commented on redundancy between Items 1 ('my child felt warmer than usual') and 2 ('my child felt hot') and suggested combining these. Two (12%) participants suggested that Item 4 ('my child had a temperature') needed to specify 'high' temperature. The above suggestions were included in Version 2 of the ObsRO after consultation with clinical experts (Table 4).

Signs of Fever
• reduced appeƟte, decreased fluid intake • lack of acƟvity, lack of social interacƟon, reduced playfulness, reduced communicaƟon • unhappy/not smiling , irritable, needy/clingy • lack of energy/lethargy, sleepy/Ɵred    14]). Participants suggested rewording a few items for clarity (e.g., 'my child interacted with family and friends' [Item 9]; participants explained that the important part is whether the child interacted, not with whom). Items were modified or deleted based on Round 1 feedback and consultation with clinical experts (Table 4).

Round 2 feedback
Thirteen participants completed, and gave feedback on, Version 2 of the ObsRO. All participants understood the instructions, although nine (52%) felt the wording could be simplified as they believed it was selfexplanatory. All participants understood the 24-h recall period and eight (62%) felt 24 h was appropriate. However, one (8%) participant felt that 24 h was too short given the duration of fever. Two (15%) felt that the 'not at all' response option was not suitable for two items ('activity level was' and 'child was quieter'). One (8%) participant would have liked to give a more detailed account of his/her child's fever.
Most participants understood all three Fever Signs items and felt they were important signs of fever. However, three (23%) participants would have liked a definition for 'high' temperature. There were no comments regarding redundancy between items. No further revisions were made to any Fever Signs items.
Most participants understood all 10 Fever Behaviour items and felt they were important. The majority of comments were concerning clarity of the wording in some items, for instance, three (23%) participants were confused by the word 'interacted' (Item 7). Three (23%) participants felt that asking whether the child was more/less irritable in the past 24 h (Item 15) was not very important as a child could be irritable for reasons other than fever. No further revisions were made to any Fever Behaviour items.
Feedback on the usability of the ObsRO instrument Most participants (n = 26, 87%) considered their overall experience of completing the ObsRO to be good. Twenty-two (73%) participants felt the ObsRO was easy to complete using their electronic device and 28 (93%) gave positive feedback about the layout of the ObsRO on their electronic device; one participant liked that all the text was on one page. Twenty-three (77%) participants found it 'easy' to navigate between questions. One participant experienced some difficulty navigating between screens as the ObsRO did not load correctly on his/her device. All participants reported they would be willing to complete this questionnaire on a daily basis during a fever episode. Participants' suggestions include adding graphics, reformatting questions so that each question stands out clearly and simplifying the date field.

Fever signs and behaviours conceptual framework
A preliminary fever signs and behaviours conceptual framework was drafted based on concept elicitation interviews. Additional modifications were made based on feedback from the cognitive interviews. The revised hypothesized conceptual framework for fever distress is shown in Fig. 4.

Discussion
The Fever Signs and Fever Behaviour ObsRO measures were developed based on a review of the NICE guidance on fever signs, input from clinical experts and based on concept elicitation and cognitive interviews with caregiver of young children < 6 years of age. In particular, the aim of the study to develop a tool to assess how to measure children's fever distress in order to enable parents and caregivers to decide on if, when and how frequently treatment with over-the-counter (OTC) medicines should be given, and to facilitate interactions with healthcare practitioners, such as community pharmacists and family doctors. The final, userfriendly ObsRO instrument comprises 13 items. These are 3 items intended as "screeners", which reflect "Fever Signs", i.e.: feeling warmer/hotter than usual; red/flushed skin; high temperature. If the parent / caregiver answers "yes" to these questions for their child, then the next 10 items pertaining to "Fever Behaviour", with questions about the child's appetite, energy levels, irritability and attention-seeking (Supplementary Material 3). The study also reported on the development of a conceptual framework to support the further development of this ObsRO.
One potential limitation of this study was that the participants were all based in the UK, so findings may not be applicable to other countries due to possible language barriers and differences in how fever is treated. The sample population was reasonably well educated; 72.0% of participants in the elicitation interviews and 56.7% of participants in the cognitive interviews had a university degree. Another potential limitation to note is that 44% of the participants had attended their family practice for their child's fever (Concept Elicitation phase). This figure is higher (25%) than others reported in the literature [2] may reflect a number of factors. However, it should be pointed out that this may also be an artefact of the small sample size (N = 11). Finally, the number of other children in the household was not recorded as part of the recruitment process. The presence of other children, particularly older siblings may be of influence parent and caregiver attitudes towards fever, and therefore we are unable to conclude from this study whether this formed a factor in the concept elicitation stage of the study.
Although there are national guidelines [10] recommending home management with over-the-counter medicines alone in the first instance, children are often admitted to hospital, yet discharged with no further medical intervention causing yet further distress to both child and their parent/ caregiver [17,18]. Fever is common in pre-school children potentially accounting for a significant degree of interactions with healthcare providers in both primary and secondary care [2,[7][8][9]. Despite often being transient in nature it leads to distress in the child, parent/caregiver and even healthcare practitioners [19]. The resulting healthcare resource utilisation clearly has an economic impact, and in addition to this the extended observation periods, additional tests and antibiotic prescription entailed through diagnostic uncertainty surrounding paediatric fever will also incur costs (with estimates at around £1000/child for infants < 3 months attending an emergency department) [20].
The causes of this understandable parental and caregiver anxiety around paediatric fever may be found in a number of factors including a lack of knowledge about the fever and appropriate fever management [21], as well as common (treatment) misconceptions and sociodemographic factors [22]. Parents and caregivers have highlighted the need for reliable and accessible information about paediatric fevers with decisions points outlines along the way [23].
Therefore, alongside educational programmes and information aimed at helping parent/caregivers to manage fever in their child, tools are needed to help parent/caregivers assess the level of distress being experienced by their child to signpost at what stage intervention is required and what level of intervention. The fever ObsRO described in this study could constitute an important toolkit in this process.

Conclusion
Results from this qualitative study provide support for the content validity of the Fever ObsRO among caregivers with children ages 3 months to 6 years. Overall, participants provided positive feedback on the Fever ObsRO and found the measure to be relevant, straightforward, easy to understand, and comprehensive of their child's fever experience. As a next step, a validation study exploring measurement properties of the Fever ObsRO in a larger sample of parents/ caregivers to ensure that the final Fever ObsRO is a clinically relevant and psychometrically sound measure of fever distress for use in young children with fever for future as an  Fig. 4 Revised preliminary fever distress conceptual framework information tool for parents and caregivers, as well as for potential use in clinical trials.