Primary care quality indicators for children: measuring quality in UK general practice

Background Child health care is an important part of the UK general practice workload; in 2009 children aged <15 years accounted for 10.9% of consultations. However, only 1.2% of the UK’s Quality and Outcomes Framework pay-for-performance incentive points relate specifically to children. Aim To improve the quality of care provided for children and adolescents by defining a set of quality indicators that reflect evidence-based national guidelines and are feasible to audit using routine computerised clinical records. Design and setting Multi-step consensus methodology in UK general practice. Method Four-step development process: selection of priority issues (applying nominal group methodology), systematic review of National Institute for Health and Care Excellence (NICE) and Scottish Intercollegiate Guidelines Network (SIGN) clinical guidelines, translation of guideline recommendations into quality indicators, and assessment of their validity and implementation feasibility (applying consensus methodology used in selecting QOF indicators). Results Of the 296 national guidelines published, 48 were potentially relevant to children in primary care, but only 123 of 1863 recommendations (6.6%) met selection criteria for translation into 56 potential quality indicators. A further 13 potential indicators were articulated after review of existing quality indicators and standards. Assessment of the validity and feasibility of implementation of these 69 candidate indicators by a clinical expert group identified 35 with median scores 8 on a 9-point Likert scale. However, only seven of the 35 achieved a GRADE rating >1 (were based on more than expert opinion). Conclusion Producing valid primary care quality indicators for children is feasible but difficult. These indicators require piloting before wide adoption but have the potential to raise the standard of primary care for all children.

funding linked to the QOF is diminishing, 5 the importance of indicators on which to base audit and clinical governance has not diminished. The consequences of low-quality care of children are reflected in unplanned hospital admissions 6 and preventable deaths. 7 In the UK, the National Institute for Health and Care Excellence (NICE) has developed an electronic library of quality standards including a number that are child specific, focused on topics such as cancer and urinary tract infections. 8 Quality indicators, measurable aspects of care against which quality standards can be set and audited, have been developed for some aspects of primary care in the US, Canada, and the Netherlands. [9][10][11][12][13] A group in the UK has proposed quality indicators for the primary care of child and adolescent mental health, 14 and indicators potentially relevant to the primary care of children can be found for a number of other disease-specific areas (such as diabetes). 15 However, there are obvious gaps. This current research attempts to develop a comprehensive set of quality indicators for the primary care of children and adolescents, which include all aspects of paediatric care and reflect existing evidence-based national guidelines.

METHOD
A multi-step approach was adopted to develop a comprehensive set of quality indicators for the primary care of children and adolescents aged <18 years. The four steps in the development process are summarised in Box 1.
The areas and aspects of care of highest priority for quality indicator development were identified by an expert panel of clinicians using the nominal group technique, 16 a structured process that has been used by others in the development of measures of care quality. 11 The membership of the group and exact method applied have been described in detail elsewhere. 17 In brief, it included UK GPs with substantial clinical experience and an interest in child health; they had a range of expertise from

Aim
To improve the quality of care provided for children and adolescents by defining a set of quality indicators that reflect evidence-based national guidelines and are feasible to audit using routine computerised clinical records.

Design and setting
Multi-step consensus methodology in UK general practice.

Method
Four-step development process: selection of priority issues (applying nominal group methodology), systematic review of National Institute for Health and Care Excellence (NICE) and Scottish Intercollegiate Guidelines Network (SIGN) clinical guidelines, translation of guideline recommendations into quality indicators, and assessment of their validity and implementation feasibility (applying consensus methodology used in selecting QOF indicators).

Results
Of the 296 national guidelines published, 48 were potentially relevant to children in primary care, but only 123 of 1863 recommendations (6.6%) met selection criteria for translation into 56 potential quality indicators. A further 13 potential indicators were articulated after review of existing quality indicators and standards. Assessment of the validity and feasibility of implementation of these 69 candidate indicators by a clinical expert group identified 35 with median scores ≥8 on a 9-point Likert scale. However, only seven of the 35 achieved a GRADE rating >1 (were based on more than expert opinion).

Conclusion
Producing valid primary care quality indicators for children is feasible but difficult. These indicators require piloting before wide adoption but have the potential to raise the standard of primary care for all children.
Keywords clinical guidelines; paediatrics; primary health care; quality indicator. commissioning, child protection, medical education and training, academic general practice, to NICE guideline development.
All published NICE and SIGN guidelines were systematically searched to identify recommendations relevant to children and adolescents. The assessment of these recommendations (against the five criteria listed in Box 1) to evaluate their potential to be quality indicators employed a modified version of the iterated consensus rating procedure (which is recommended as the optimal process to identify recommendations with quality indicator development potential). 18,19 The final clinical assessment of feasibility and validity was based on a method devised by the RAND organisation (the most widely used methodology to generate professional consensus on quality indicators). 20,21 The panel (n = 13, 10 at final meeting) was of similar composition to that which participated in the nominal group technique rating, with some overlap in membership. It was provided with the GRADE rating for each indicator and referenced source material. 22 Panel members were required to rate the validity and feasibility of the proposed indicators on a 9-point Likert scale (1 = strongly disagree, 9 = strongly agree). Figure 1 summarises the outcome of the development process. The nominal group discussion identified six broad clinical areas in which development of quality was agreed to be important. Three focused on clinical management of illness (early recognition of serious illness, management of common conditions, and mental health) and three on more generic aspects of child care (child protection, health promotion, and governance). The governance issues ranged from illness specific (for example, unnecessary admission of children to hospital with acute asthma) to standards of professional development (for example, review of paediatric-specific skills in annual appraisal).

National guideline review
Review of the 296 national guidelines identified 48 (37 NICE and 11 SIGN) relevant to children or adolescents, containing 1863 potentially relevant recommendations. However, only 123 (6.6%) met all five selection criteria and these were taken forward for translation into quality indicators. Most of these recommendations were about clinical management of illness (33 early recognition of serious illness; 51 management of common conditions; 28 mental health). Only three recommendations selected were about child protection, eight about health promotion, and none about governance.

Articulation and assessment of indicators
The 123 recommendations were translated into 56 quality indicators; this reduction mainly reflected overlap (that is, more than one recommendation could be addressed

How this fits in
Child health care is an important part of general practice workload in the UK, but is very poorly represented in the financially incentivised Quality and Outcomes Framework. Although NICE has developed a small number of relevant quality standards, primary care quality indicators have been developed for specific conditions, and indicator sets for paediatric ambulatory care have been developed in North America and Holland, there remains an important gap. This current research has developed a comprehensive set of quality indicators relevant to the care of children and adolescents. The indicators could be feasibly implemented in the UK by audit of computerised general practice records.
British Journal of General Practice, December 2014 e753

Box 1. Summary of method for developing quality indicators
Step 1: Selection of priority areas for quality indicator development Step 2: Identification of relevant recommendations from national guidelines Step 3: Translation of selected guideline recommendations into quality indicators Step 4: Final assessment of quality indicators for feasibility and reliability The clinical areas for which priority quality indicators are most needed were identified by an expert review panel using the nominal group technique; the expert panel was informed by evidence obtained by qualitative interviews with practising primary care clinicians, evidence from a systematic review of effective clinical interventions, and an analysis of unplanned hospital admissions for primary care-sensitive conditions.
All national guidelines for England (NICE) and Scotland (SIGN) were reviewed to select those potentially relevant to children and dealing with issues in the prioritised clinical areas. All recommendations were extracted but were selected for further development only if they: 1) made a precise statement about what constitutes high-quality care; 2) defined a standard against which care quality could be measured; 3) were clearly defined; 4) were measurable; and 5) were attributable to actions in primary care.
To inform the drafting process, exemplar quality indicators were identified from three specific sources: 1) the Agency for Healthcare Research and Quality (AHRQ) National Quality Measures Clearinghouse website; 2) the Royal College of General Practitioners (RCGP) Training Standards and other previously proposed quality indicators for UK general practice; 3) paediatric indicators previously developed by RAND. If the above sources contributed no useful exemplars, PubMed was searched for newly published indicators.
All the quality indicators were further assessed by a second expert review panel (with some overlap in membership with the panel that undertook Step 1) using the RAND appropriateness method. by a single indicator). However, the process of identifying exemplar indicators from other sources not only facilitated drafting but also identified indicators for the generic aspects of clinical care not covered by guideline recommendations. A further 11 indicators were added through this mechanism (seven relating to governance, two each relating to health promotion and child protection) and two articulated by panel members during the final assessment phase. A total of 69 potential quality indicators were therefore included in the final assessment of feasibility and validity, of which 35 achieved a median score of ≥8. Of these 35 indicators, 28 were rated as GRADE 1 (expert consensus) while four were rated 2 (low level of evidence), one was rated 3 (moderate level of evidence) and two were rated 4 (high level of evidence).

Quality indicators for routine care
Box 2 shows the indicators relating to the quality of routine clinical care. Most cover common conditions seen mainly in preschool children including acute infection, acute diarrhoea, and allergic reactions. Four indicators relate to more specific age groups: neonatal jaundice, infant colic, and adolescent self-harm. Of the three indicators that are not focused on specific conditions, two are process measures (professional development and the availability of equipment) but one is a health service outcome measure: the number of accident and emergency attendances by children registered with the practice. The panel recognised that this would need to be standardised for other factors such as the age profile of the child population and the practice deprivation index as both will influence accident and emergency attendance rates.

Quality indicators for chronic illness
Box 3 shows the indicators relating to the quality of recognition and care of chronic illness. Not surprisingly, four are about asthma (the most common chronic disease managed in primary care). The other specific chronic conditions covered were eczema, coeliac disease, diabetes, mental health, epilepsy, hepatitis B risk, and attention deficit hyperactivity disorder (ADHD). The one generic indicator is about long-term prescribing.

Quality indicators for child protection and developmental assessment
Box 4 lists the indicators selected for child protection and developmental assessment. One indicator, the taking of appropriate urgent action on new-onset fixed squint, is a developmental issue but could equally be seen as an important indicator of highquality routine care. The other indicators focus on safeguarding; two on professional development and training, and three on

DISCUSSION Summary
Producing quality indicators is difficult. It took 2 years to produce 35 indicators that have high levels of clinical support and reflect national guidelines. These indicators were developed by a multi-step method very similar to that used in formulating the QOF. They cover a range of clinical issues agreed to be of importance to high-quality primary care for children and adolescents. However, there are many important aspects of care that are not covered and most indicators are supported only by low-level evidence (mainly expert opinion). They need to be piloted before large scale implementation to check for unforeseen difficulties in use and adverse consequences but they do fill an important gap. Lack of a set of quality indicators is no longer an excuse for the omission of markers of paediatric care from the QOF.

Strengths and limitations
The main strength of the research undertaken was that it followed a welldefined multi-step process that has been developed and refined by others. As stated in the methods, the nominal group technique was used by Guttmann et al, in defining important clinical areas in which to develop quality indicators for emergency paediatric care in Canada. 11 The RAND appropriateness method, developed in North America, is now a core feature of the process of development of quality indicators for the UK QOF. 21,22 The initial assessment of the most important areas in which to develop quality indicators was informed by research (a qualitative survey of service GPs 23 and an analysis of emergency hospital admissions). 6 Every attempt was made to ensure the indicators were evidence based (by basing the initial drafting on national clinical guidelines prepared by NICE).
The main weakness of the process was that the only stakeholders consulted were GPs. This approach was taken because the initial implementation concern was professional acceptability. In the UK, paediatric care is provided by the whole primary care team, with health visitors in particular playing an important role in promoting child health, safeguarding children, and supporting families of children aged <5 years. If the process was repeated it would involve other primary care clinicians (for example nurses or health visitors) in the expert panels and consultation with other stakeholders (for example, parents, hospital and community paediatricians, or health service commissioners). Limiting the process to clinicians with a professional and financial interest in the outcome has the obvious potential to influence both the areas chosen as important for development and the assessment of validity and feasibility. The study departed from a strict application of the consensus method by including indicators in the final set based on their median rating for validity and implementation even if complete consensus was not achieved. The authors believe that seeking complete consensus risks a 'lowest common denominator' effect, with the exclusion of otherwise highly-rated indicators because they were opposed by one or two individuals with a strong contrary opinion. For example, 'children with atopic eczema should be prescribed emollients' achieved median ratings of 8 and 9 but would have been rejected because one panel member gave a 2 rating for both feasibility and validity.

Comparison with existing literature
Most of the previously developed paediatric quality indicators focus on care of acute illness (for example, diabetic ketoacidosis, status asthmaticus, anaphylaxis, status epilepticus, severe head injury, and sepsis). 24 The quality indicators used to assess the care of children in the US Medicaid programme looked at primary care but focused on vaccination, access, and procedures. 25 The set of paediatric quality indicators developed by Giesen in the Netherlands also focused on primary care and used a very similar approach of developing indicators based on national guidelines but focused only on prescribing and referral out-of-hours. 12 The most inclusive set of primary care quality indicators for children and adolescents was developed in the US by Mangione-Smith et al in 2007 using the RAND methodology, but it involves 175 indicators (covering management of acne, ADHD, allergic rhinitis, asthma, depression, diarrhoea, fever, urinary tract infection, and sexually transmitted Infections as well as prevention and care of well children) and is more useful for one-off surveys of care quality rather than routine monitoring and quality assurance. 10 There is some overlap between our work and the quality standards that NICE are starting to publish, some of which apply to children. 8 For example, the NICE standard for primary care of epilepsy is that 'Children and young people presenting with a suspected seizure are seen by a specialist in the diagnosis and management of the epilepsies within 2 weeks of presentation.' This study's comparative indicator is formulated more precisely in terms of the children who needed to be referred but it is less precise on timing: 'Children with a first non-febrile seizure should have clearly documented evidence of referral to secondary care for further assessment.' 26 The varying language in which our indicators are expressed highlights the overlap between the concepts of a quality indicator and quality standard. Some are expressed as true indicators without any predefined standard (for example, accident and emergency attendances for children in previous 12-month period) while others imply a zero-tolerance standard (for example, infants with colic should not be prescribed dicyclomine). In order to apply the indicators in clinical practice it is necessary to set a standard (that is, the level which must be achieved) but it should be noted that QOF indicators seldom set a zero-tolerance standard: full points are awarded for less than 100% compliance.

Implications for research and practice
The most striking feature of the selected indicators is their low GRADE scores: the evidence base appears weak. This is in contrast to the reported evidence base for the quality of acute hospital-based paediatric care reported by Stang et al where 37% per cent of the selected indicators are based on moderate or high-quality evidence. 24 To some extent this must reflect a lack of evidence about the optimal way of providing high-quality primary care to children; it also reflects the failure to select indicators from areas of care where the evidence is strongest (for example, vaccination). And perhaps it also reflects the nature of much of the existing evidence base with its focus on the absolute effectiveness of drugs and procedures rather than the optimal mechanism of their use (for example, when, to whom, and how much medicine is given).
Experience of the QOF in the UK has highlighted the importance of pilot-testing quality indicators before wide-scale national roll-out. 21 The target population has to be carefully defined, precise coding developed to extract reliable information from computerised records, and the potential for gaming and the generation of unexpected adverse outcomes assessed. But it is important that care quality for children and adolescents is not ignored. Also, it is important that auditable quality indicators cover the full range of care. The indicator set that has been developed here reflects professional concerns and existing guidelines. In the UK it would be implementable through computerised audit of existing primary care records, whether or not it was used just as an audit tool or linked to QOF payment by results. So while the indicator set is not perfect, it is a start and improvement is most likely if it is taken up and used. The funding bodies had no role in the study design; collection, analysis, interpretation of data; in writing the report; nor decision to submit the article for publication. The views expressed in this paper are those of the authors and not necessarily those of the funding bodies.

Ethical approval
The Proportionate Review Sub-Committee of the South West London 4 Research Ethics Committee reviewed the project (11/H0806/6) and deemed it service development, not requiring approval from a Research Ethics Committee.

Provenance
Freely submitted; externally peer reviewed.