A pilot study to develop assessment tools for Group A Streptococcus surveillance studies

Ethics and study setting

This study took place at an Aboriginal Community Controlled Health Organization primary health care clinic in Broome, the largest town of the Kimberley, Western Australia (WA) in September and October 2017. Local community and key stakeholders were consulted over the preceding 6 months. This included consultation with clinic staff to ensure the study was a shared priority and could be conducted concurrently with clinical activities. Letters of support were obtained from the clinic and the Kimberley Aboriginal Health Planning Forum Research Subcommittee (KAHPF, 2017-008). Ethics approval was granted by the Western Australian Aboriginal Health Ethics Committee (Approval 782) and University of Western Australia Human Research Ethics Office (RA/4/1/9206).

Yarning circle to develop culturally informed sore throat checklist

Over five days in September and October 2017, caregivers of and health care workers for Aboriginal children were invited to participate in a yarning circle to ascertain the preferred words and stories used to describe sore throats in children living in the Broome region of Western Australia. Study staff obtained informed consent and performed semi-structured interviews with open ended questions and symptom conversation triggers (Table 1). A qualitative researcher with decades of experience using the methodology of yarning (Bessarab & Ng’andu, 2010) with Aboriginal families informed the semi-structured interview guide and trained study staff. The yarning circle was led by a non-Aboriginal researcher accompanied by an Aboriginal health care worker who had recruited participants for the research. Yarnings were recorded with permission and transcribed verbatim. Recruitment of participants ceased when data saturation was achieved. Each transcript was reviewed individually; the breadth of sore throat symptoms was collated from all interviews and grouped across eight contextual categories (sore throat stories, self-description, local words, child description, awareness, spontaneously mentioned symptom, checklist reported symptom, checklist noticed signs). The most frequently described sore throat symptoms (Figs. 1A–1H) informed the finalized list of questions for the ‘sore throat story checklist’ for use in the cross-sectional pilot surveillance study (Table 2).

Study pilot and participants

The cross-sectional pilot surveillance study took place in the same clinic setting as the yarning circles and was performed over two days in October 2017. Caregivers of participants were approached in the waiting room. The primary purpose for their attendance was not recorded but could be to access healthcare for children or adults. Caregivers of Aboriginal children aged 3–15 years provided written informed consent for their child/children; children ≥7 years provided written assent. Children with immune deficiency and known diagnosis of RHD and who were receiving secondary prophylaxis with benzathine penicillin G (BPG) were excluded. A sample of 15 participants was considered achievable to recruit and assess over 2 days. With their caregiver present, children were interviewed and assessed by a student doctor and a pediatrician. A clinical reference form that collected data was used for data collection. Demographic information (age, Aboriginal status, gender, mean number of people and bedrooms in household, eligibility, and school attendance) were recorded. To assess for GAS impetigo and pharyngitis, research study staff performed: (1) a checklist of questions to ascertain current sore throat and skin symptoms, (2) clinical assessments, (3) photographs, (4) bacteriological tests, and (5) serological tests, which are detailed (Table 3). All children were seen on the same day by their usual care clinician and offered treatment, where indicated, for pharyngitis or impetigo. Data were recorded in the web-based Research Electronic Data Capture (REDCap) platform (Harris et al., 2009) (Vanderbilt, Nashville, USA) during the assessment.

Symptom data collection and examination

Children were screened for pharyngitis using the checklist of six questions (Table 2) that were developed from the sore throat yarning sessions. The Modified Centor Criteria (McIsaac et al., 2000), a validated tool for throat examination, was used to determine the risk of GAS pharyngitis. Tonsillar size was assessed by the research study clinician using the Brodsky scale (Brodsky, 1989), a standardized and reproducible measurement. The skin symptom checklist was previously validated in a surveillance program and clinical trial and used without change (Bowen et al., 2014c; Andrews et al., 2009). Examination of the visible skin recorded purulent, crusted and flat/dry sores on a body map using a standardized tool (Recognition and Treatment of Skin Infections protocol) (Menzies School of Health Research, 2009). When a child indicated a skin sore under their clothes, these sores were examined and documented.

Photography

An iPhone 7 Plus (Apple Inc., Cupertino, USA) mobile device was used to conduct all digital photography of the pharynx and skin. Approaches to pharynx photography were developed into a standard operating procedure with the assistance of the Perth Children’s Hospital (PCH) otolaryngologists and hospital photographer (Sampson, Mullane & Bowen, 2017). Photographs of the pharynx and skin used standardized equipment and camera settings, participant positioning and photography technique. Photographs were then uploaded to REDCap for objective grading of the tonsils using the Brodsky Grading Scale (Brodsky, 1989) by two independent pediatricians. This independent assessment aimed to provide data on the feasibility and accuracy of using photographs to help assess sore throats in other clinical applications, such as in remote and telehealth contexts. For each participant, the most severe skin sore was photographed using an iPhone adapted version of the photographic protocol described by Bowen et al. (2014c)

Bacteriological tests

Throats were swabbed simultaneously for GAS Rapid Antigen Detection Testing (RADT) (Stewart et al., 2014) and culture by holding two sterile swab sticks together while passing them over the tonsils and avoiding the tongue. The RADT (Alere BinaxNow GAS, Abbot, Chicago, USA) was performed immediately according to manufacturer’s instructions. The most severe skin sore was swabbed as previously described. (Bowen et al., 2014c; Bowen et al., 2014d) Throat and skin swabs were placed in vials of skim milk tryptone glucose glycogen broth (STGGB, PathWest Media, Perth, Australia), and kept at 4–8 °C until culture at the PathWest diagnostic laboratory, Broome, WA. GAS were identified (Andrews et al., 2009) from 10 µL of STGGB streaked on horse blood agar plates and incubated for 24 h at 35–37 °C. Suspected isolates were confirmed with a latex streptococcal grouping kit (Oxoid, Basingstoke, United Kingdom). The common skin pathogen, Staphylococcus aureus, was also identified from skin swabs with positive slide (Oxoid) and tube coagulase results, with methicillin susceptibility confirmed by the cefoxitin disc diffusion method (FOX 30, Oxoid, United Kingdom) (Bowen et al., 2013).

Serological tests

Blood was collected from a single fingerprick sample; blood droplets squeezed onto Whatman 903 filter paper (Interpath, Perth, Western Australia) to fill all five circles. The resulting dried blood spot (DBS) samples were air dried at room temperature for 24 h and then stored with desiccant at 4–8 °C for transportation to Perth, WA. Two blood spot circles were independently processed per card to detect anti-streptolysin O (ASO) antibodies at PathWest Laboratories (Perth, Australia) on a nephelometer as previously described (Joseph et al., 2017).

Feasibility, acceptability, and budgeting for scaled surveillance

Throughout the study, participants and caregivers were asked about their comfort, willingness to continue and whether they had any questions. Feedback from participants was sought to determine the acceptability of the consenting process and their experience with each aspect of data collection in field notes by study staff about the questionnaires, clinical checks, throat skin and blood sampling and photography) during their participation. Study notes of participant feedback and the ease or difficulty of performing each assessment tool were captured by study staff and compiled as raw data at the end of the pilot. To inform future scaling of throat and skin surveillance, study staff calculated the time taken to perform each aspect of the assessment, staffing and consumable costs. The total cost for the surveillance pilot was calculated, and an average cost per child determined.

Data analysis

Statistical analysis of descriptive and inferential statistics used GraphPad Prism 9.1.2 (GraphPad Software Inc, San Diego, CA, USA).

Sore throat yarning reveals key sore throat descriptors

Sore throat yarning achieved data saturation with ten Aboriginal participants. All were mothers (five aged 25–39 years, five aged 40–59 years) that had additional caring roles to other children including as grandmothers (n = 5), aunts (n = 10), and carers or guardians (n = 7). Stories about caring for children with sore throats identified nine symptoms, whereas self-descriptors identified 12 symptoms (Figs. 1A, 1B). ‘Sore throat,’ and ‘hard to swallow’ were the predominant symptoms. No local words were used by participants to describe a sore throat. Additional descriptors were identified by ‘those used by children’ (Fig. 1C) and included ‘cough’, ‘touching throat’, ‘not eating’, and ‘not talking’, however soreness and hard to swallow remained the predominant descriptors. Other signs and symptoms revealed less commonly reported symptoms (Fig. 1D) with feeling hot/cold and weak/run down predominating. Thirty-two symptoms were recorded from conversational prompts (Fig. 1E). One participant identified there was a possibility of a child not reporting a sore throat “so that [child] could keep playing.” The remaining participants reported high awareness for sore throat illness in children they cared for. Pre-defined checklist prompts for ‘reported’ and ‘noticed’ symptoms (Figs. 1F and 1G) identified 22 symptoms, including additional descriptors for the sore throat story (croaky voice, achy ears, bad breath). Croaky voice was an addition to the ‘predominant descriptors’ that were first identified in conversations (hurts when swallowing, not eating/drinking, sore, feeling hot/cold). Six predominant descriptors were chosen for the sore throat checklist (Table 2). Despite being recorded as prevalently in some instances, the descriptors, ‘tired,’ ‘weak’ and ‘run-down’ were not considered specific for sore throat and therefore excluded from the developed checklist.

Symptom data collection and examination

Children were assessed in family groups of between 1–5 children at a time. Of the 21 children approached, 17 were eligible and recruited (mean age 9 years [interquartile range (IQR) 5.5–13.5], 65% female). Of the 17 children, three declined the finger-prick test and one the throat swab. All other elements were completed by all children.

Seven participants answered affirmatively to one or more of the sore throat symptoms questions (Table 2), in two of whom the sole symptom ‘feeling hot or cold’, five reported ‘throat pain’, four ‘hard to swallow’, two ‘not eating’, two ‘croaky voice’ and six ‘feeling hot or cold’. ‘Not drinking’ was the only sore throat descriptor not affirmatively reported in this pilot. Nine (53%) participants had a Modified Centor Criteria score >1 on examination, with 3/17 declared to have had prior tonsillectomy. Clinical examination identified two children with a Modified Centor score of 2 (due to age and absence of cough), however as these children were asymptomatic, this score was not clinically relevant. Consequently, appropriate clinical examination did not identify sore throats in addition to those reported by children using the sore throat story.

Skin infection symptoms were reported by 5/17 (29%) participants with a reported duration of <1 day (n = 1), 1–3 days (n = 2), 3–5 days (n = 1) and >5 days (n = 1). Four reported these sores were itchy, whilst three reported that they were painful/uncomfortable. On clinical examination, each of these children were recognized to have current impetigo lesions (crusted and purulent). Clinical examination identified an additional ten children with skin sores. Of the 15 total children with skin sores, 11 presented with a combination of ‘purulent and crusted’ or ‘purulent and flat/dry’ impetigo. Two children presented with scabies. The total number of sores ranged from 0–42 (mean 7.5, median three [IQR 1.5−9.5]).

Three children reported sore throat and skin sores concurrently, four a sore throat only and two skin sores only. Clinical examination identified concurrent sore throat and skin sores in seven children, a further two children with skin sores only, and no children with sore throats only.

Photography

All participants consented to throat photography (Fig. 2). Two independent clinicians assigned Brodsky grading scales for all pharynx images that were deemed good quality (15/17). Study notes recorded the two instances of difficultly with photography were due to a reluctance for one participant to open their mouth wide for photography, and a small oral cavity obscuring the pharynx in another. On clinical assessment, the median Brodsky grading score was two, consistent with the median Brodsky grading score from photographs of 1.5. The clinically assigned score on the day and that assigned via photograph analysis was only equivalent on five of 15 occasions (grade 0 = twice, grade 1 twice, grade 3 once). Of those that differed, nine were a single score apart, and one was two scores apart. Photography captured photo evidence of tonsillectomy in two children, with one also having evidence of pharyngitis.

Bacteriological and serological testing

All RADTs were negative (n = 16). Throat swabs from 4 participants identified scanty GAS, scored as 1+ out of a possible 4+ for each of the plates. All four children who had GAS cultured from throat swabs were referred for treatment with intramuscular benzyl benzathine penicillin as a conservative approach to prevent RF. Of these four, only one presented with sore throat symptoms (throat pain, croaky voice and fever) for which she had attend clinic for care on the same day. Four additional participants cultured group C Streptococcus from throat swabs of which two presented with sore throat symptoms –none were referred for treatment. Fifteen participants had skin sores swabbed. Of the 15 sore swabs, one cultured GAS and this same participant had GAS in the throat with symptomatic sore throat presentation. Three participants cultured S. aureus from skin swabs of which one was identified as methicillin resistant (MRSA). GAS and S. aureus co-infection was found in 1/15 sores. Of 14 participants consenting to DBS, the median anti-Streptolysin O titer (ASOT) was 306 IU/mL ([IQR] 220.3–776.8). Most ASOT values (12/14) exceeded the upper limit of normal (<200IU/mL) (Steer et al., 2009).

Feasibility and acceptability

Acceptability and engagement were inferred from study notes, the positive responses, high but not universal consent rates, engagement in activities and low rates of refusal to any aspect of the pilot study. The participating children were very engaged in the assessments, had high uptake for the more invasive procedures e.g., throat swab (16/17) and finger prick (14/17) and enjoyed assisting with the RADT processing as they were able to understand the science of why their throats were being swabbed. Participants and their families asked appropriate knowledge seeking questions throughout the interaction. Although three children did not consent to the finger-prick DBS, this was reported by the carer to coincide with historical resistance to blood draws.

The total mean time for screening was 19 min per child (standard deviation [SD] ±5.6 min), with a mean of 23 min on day one and 16 min on day two. The fastest assessment in participant 16 of 17 was 11 min, with two study staff collecting data. It was noted that confidence with the surveillance activities improved by the research team with progressively shorter assessment times. Surveillance costs per child was estimated to be $89.63 AUD. Other estimated costs of the study, including once-off costs were recorded (Table 4). A flow diagram indicating steps in study and major findings is presented (Fig. 3).