National Diabetes Inpatient Safety Audit (NDISA) 2018-2021

The National Diabetes Inpatient Safety Audit (NDISA) is part of the National Diabetes Audit (NDA) portfolio within the National Clinical Audit and Patient Outcomes Programme (NCAPOP), commissioned by the Healthcare Quality Improvement Partnership (HQIP) and funded by NHS England and the Welsh Government.

The NDISA has 2 main strands:

Firstly, NDISA undertakes a continuous collection of serious inpatient harms that can affect people with diabetes. This collection was previously known as the National Diabetes Inpatient Audit (NaDIA) harms audit. Data collection began on 1 May 2018. All acute hospitals in England with inpatients with diabetes are eligible to participate. The Welsh government decided not to participate in the NDISA harms collection.

Secondly NDISA reviews the provision of inpatient services using the Integrated Specialist Services Structures Survey (ISSSS, abbreviated to ISS), an NDA questionnaire which asks healthcare providers whether they have put in place structures and systems of care for people with diabetes recommended by the Getting It Right First Time (GIRFT) programme. The ISS was distributed to healthcare providers in England and Wales in October 2021.

The NDISA collects data on occurrences of the following serious inpatient harms that can affect people with diabetes:

• hypoglycaemia requiring rescue treatment
• diabetic ketoacidosis (DKA)
• hyperglycaemic hyperosmolar state (HHS)
• new diabetic foot ulceration (DFU)

The main objective of the NDISA harms collection is to help reduce the rates of these serious inpatient harms by providing case-mix adjusted benchmarked feedback and identifying patients at risk to hospital trusts and inform quality improvement work. Monitoring and improving participation are important first steps towards meeting this objective.

The NDISA 2018-21 report includes analysis on participation, a count of the number of participants and harms, and reviews changes in the rate of inpatient harms over time. The NDISA report also includes patient profiles of each inpatient harm, which comprise of demographics, diabetes characteristics, treatment targets, care processes, admission characteristics and comorbidities. These analyses were produced through linkage to the core NDA and Hospital Episode Statistics (HES) data.

NDISA also reviews the provision of inpatient services in care providers (acute NHS trusts in England and Welsh Local Health Boards, LHBs), benchmarked against recommendations made by the GIRFT Programme National Specialty Report for Diabetes (2020). The GIRFT report was itself informed by recommendations made in previous NaDIA publications (both the NaDIA snapshot audit and the NaDIA harms collection), the NHS Long Term Plan (2019), the Diabetes UK 2018 publication on Making Hospitals Safe For People With Diabetes and the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) Report into Perioperative Diabetes (2018).

The NDISA 2018-21 report will be of interest to the public, especially to people with diabetes. Health planners and policy makers, as well as acute NHS Trusts, Clinical Commissioning Groups (CCGs), Local Health Boards (LHBs), Integrated Care Systems (ICSs), Integrated Care Boards (ICBs), Sustainability and Transformation Partnerships (STPs), Clinical Networks (CNs; formerly Strategic Clinical Networks or SCNs) and other providers and commissioners of specialist diabetes services will also make use of the information in this report.

Patients in the NDISA harms collection were linked to data items recorded in the core NDA, including NDA data up to 30 September 2021. Linkage included data from the incomplete core NDA year 2021-22 for England only. NDA data covering April to September 2021 is therefore provisional and does not include Welsh data.

The core NDA collects data on patient demographics, care processes and treatment targets amongst those registered with participating GP practices and secondary care organisations in England and Wales.

GP practice participation in England and Wales was 99.3% in the latest core NDA Report 1 for 2019-20, covering 1 January 2019 to 31 March 2020. An additional 98 specialist services in England submitted data to the collection.

94.0% of patients in the NDISA harms collection were found in the core NDA. Age at start of NDA year, sex, ethnicity, diabetes type and diabetes duration at start of NDA year were taken from the core NDA demographics table. Smoking status, body mass index (BMI), renal function and deprivation quintile took the value in NDA closest to the inpatient harm date (in the corresponding NDA year or 1 of the preceding 3 years). NDA treatment targets and care processes were taken from the NDA year preceding the inpatient harm year (defined as April to March).

HES is a database containing details of all admissions, outpatient appointments and accident and emergency attendances at NHS hospitals and NHS-funded private providers in England. NDISA harms patients were linked to hospital admissions data in HES, with admissions data available up to 31 December 2021. HES data covering April to December 2021 is provisional.

Hospital admissions covering the date of the inpatient harm were found for 94.4% of episodes in the NDISA harms collection.

A comparison population of inpatients with diabetes was created by linking people in the core NDA to hospital admissions in HES, covering the 42 months of the NDISA harms collection (May 2018 to October 2021).

Resultant comparisons between the NDISA harms collection-HES cohort and the NDA-HES cohort used occupied bed days rather than unique patients or admissions as the unit of currency, thereby adjusting for the length of the hospital stay. Where comparisons were made, occupied bed days on which a patient’s harm was reported to the NDISA harms collection were removed from the NDA-HES comparison population.

Data in the inpatient harms section of the NDISA 2018-21 report are derived from harms that occurred between 1 May 2018 and 31 October 2021 (the cohort), covering the first 42 months of data collection. The data was extracted from the NDISA harms collection database on 3 December 2021, 1 month after the end of the audit period.

Results in the inpatient service provision section of the NDISA 2018-21 report are derived from ISS data for 1 October 2021. The ISS data was extracted from the ISS data collection tool on 2 December 2021, following the close of the ISS submission window.

The NDISA 2018-21 report was published on 14 July 2022. The time lag to the publication of the main report was therefore 9.5 months after the inpatient services snapshot (1 October 2021), 8.5 months after the end of the harms cohort (31 October 2021) and 7.5 months after the respective harms and ISS data extracts were taken (early December 2021).

Participation in the NDISA harms collection is voluntary, but encouraged, for all acute NHS trusts in England admitting patients with diabetes. The audit does not collect data from outside England, so organisations and inpatients from Wales are not included.

113 NHS trusts registered for the NDISA harms collection between 1 May 2018 and 3 December 2021. 109 of these 113 NHS trusts participated in the NDISA harms collection between 1 May 2018 and 31 October 2021 (the audit period). An NHS trust is classed as having participated if they either:

1. Submitted 1 or more harm; or
2. Confirmed a null return for 1 or more month

during the audit period. 63 NHS trusts submitted on a regular basis, defined as having met criteria a) or b) above in all 8 quarterly periods in the last 2 years of the collection (November 2018 to October 2021).

For comparison, 126 NHS trusts in England are known to be eligible for the NDISA harms collection, with eligibility inferred from participation in the NaDIA snapshot collections between 2015 and 2019 or through participation in the NDISA harms collection itself. This suggests that NDISA harms participation is around 87% of applicable NHS trusts (109 of 126).

6,150 inpatient harms were recorded in the NDISA harms collection during the audit period. Case ascertainment for each type of inpatient harm has previously been calculated using estimates derived from the 2019 NaDIA snapshot. Results were published in the 2019 NaDIA harms audit report (p. 32). Case ascertainment for DKA was estimated at 20% (DKA), with the other harms estimated at 6-8% (others). Re-running the same methodology on the latest NDISA harms collection data produces similar figures (21% for DKA and 5-8% for the other harms).

However, the true case ascertainment is likely to be higher than the above estimates for 2 reasons:

1. The expected number of harms may be inflated by the increased likelihood of longer stay patients both experiencing a harm and being present on the NaDIA snapshot audit day; and
2. Because the latest NaDIA snapshot was undertaken in September 2019, the likely reduction in the number of inpatient harms since this period (see slide 9 in the NDISA 2018-21 report) will not be reflected in the baseline figures.

Both of the factors above will inflate the expected number of harms, consequently reducing the case ascertainment. It was therefore decided not to publish detailed case ascertainment results in this year’s NDISA report, though it is acknowledged the true case ascertainment is likely to be relatively low.

Analysis covering April to October 2021 uses provisional data from both HES and core NDA. Further analysis will be required to get a complete picture of hospital activity during this period.

It should be noted that different NDISA harms cohorts are used in different parts of the NDISA 2018-21 report:

• The whole NDISA harms cohort (May 2018 to October 2021) was used to review the number of harms submitted in each quarter since audit inception (Chart 4.1 and Table 4.1 of the main report).
• The whole NDISA harms cohort was also used to build the patient profiles, thereby maximising the volume of patient characteristics that can be assigned to each harm type for comparison against the baseline population of non-harms bed days (Tables 5.1-5.5 of the report).
• The review of harms rates over time (Charts 1.2 and 4.2-4.3 in the main report) starts at the beginning of the first full calendar year of the NDISA harms collection (Q1 2019), primarily to exclude the early period of the audit collection (Q2 and Q3 2018) when the number of audit participants and harms was lower than in subsequent quarters (see Table 3), leading to artificially low harms rates.

The ISS is a survey conducted at provider level (mainly NHS trust and Welsh LHBs) that looks at service providers as of 1 October each year. Whereas the NDISA harms collection covers England only, the ISS covers both English and Welsh providers. The first ISS collection undertaken for October 2020, although the questions were different to those asked in the 2021 survey.

96 providers in England and Wales responded to the 2021 ISS (see Table 1, below). The NDISA 2018-21 report includes data from 85 of the 96 ISS responders which provide acute inpatient medical care, such as active short-term hospital treatment for severe injury, illness, urgent medical conditions, or recovery from surgery. There are 138 acute providers in England¹ and 7 in Wales², so the overall ISS response rate for acute providers is 59% (85 of 145).

Table 1: Submitters to the ISS, England and Wales, October 2021

Table 2 shows that, with the exception of the first quarter of data collection (May to June 2018), the number of participating NHS trusts has been consistently in the 80s (between 81 and 88). As NHS trusts should verify monthly nil submissions, fluctuations in the underlying number of harms should not affect the number of participants.

Table 2: Number of participating NHS trusts¹, by quarter, England, May 2018 - October 2021

Footnotes

1. The number of unique NHS trusts that participated in the NDISA harms collection during the stated period. Participation is defined as 1 or more inpatient harms submission or any verified monthly nil submission in the quarterly period. The "Total" represents the total number of unique providers who made a submission in any of the quarters within the audit period.

Table 3 shows that the number of inpatient harms submitted on a quarterly basis increased from 325 in quarter 1 (May to June 2018) to a peak of 625 in quarter 3 (November 2018-January 2019). The number of harms has subsequently been between 325 and 500 per quarter, with a general downward trend. It is not known to what extent this trend reflects changes in NDISA harms collection participation or real changes in the underlying occurrence of harms (e.g. seasonal changes), with 2020 and 2021 figures likely to be affected by the COVID-19 pandemic.

Table 3: Number of inpatient harms, by harm type and quarter when inpatient harm occurred, England, May 2018 - October 2020 (rounded¹)

Footnotes

1. Counts have been rounded. Counts between 1 and 7 are represented as a 5. All counts greater than 7 have been rounded to the nearest 5. Consequently, the total will not usually match the sum of the 4 constituent inpatient harms.

2. As noted above, an earlier ISS collection was undertaken in October 2020, although the questions were different to those asked in the 2021, so the results are not comparable.

From 2010 to 2019 the NaDIA snapshot audit collected information on the incidence of patient harms during a specified week in September, including the 4 patient harms collected in the continuous NDISA harms collection. The latest NaDIA snapshot was carried out by hospital teams in England on a nominated day between 23 and 27 September 2019 and published in the 2019 NaDIA report published on 13 November 2020.

Patient harms in the NaDIA snapshot are reported as a proportion of inpatients experiencing the harm during their hospital stay (diabetic ketoacidosis (DKA), hyperglycaemic hyperosmolar state (HHS), new diabetic foot ulceration (DFU)) or in the previous 7 days of their hospital stay (hypoglycaemia requiring rescue treatment).

Direct comparison between the collections is difficult due to the different methodology and collection periods of the 2 collections. Comparative analysis was done for the 2020 NaDIA harms report to produce estimated case ascertainment for each inpatient harm. The results and limitations of this comparison are discussed above (in Accuracy, reliability and limitations).

No other data source provides equivalent information about the incidence of inpatient harms in England.

From 2010 to 2019 the NaDIA snapshot ran a Hospital Characteristics (HC) survey as part of the annual NaDIA snapshot undertaken in September, providing information on the hospital’s resources and staffing structure. The latest results from the NaDIA HC survey were published as part of the 2019 NaDIA report.

However, direct comparison with the care structures captured in the 2021 ISS collection is not possible. Firstly, the questions in the NaDIA HC are different from those in the 2021 ISS collection. Where there are similarities in subject matter, the impact of differences in wording should be considered before comparing. For example, the questions below around the provision of specialist foot teams are superficially similar, but the exact wording needs to be reviewed carefully before comparisons are made:

• 2019 NaDIA report: Is there an established Multi-disciplinary Diabetic Foot care Team (MDFT)?
• 2021 ISS collection: Does your Trust (England) or Local Health Board (Wales) have a dedicated multi-disciplinary foot care service (MDFS) as stated in the NHS Long Term Plan and NICE NG19?

A secondary issue to consider is that the units used in the NaDIA HC (a mixture of hospitals, multi-hospital groupings and providers) differ from those used in the ISS (provider), again presenting an obstacle to comparability.

Audit information is held securely and with restricted access.

It is expected that, through the audit collection, all organisations will continue to follow existing NHS codes of practice about patient confidentiality, information security management, record management and other legal obligations.

A risk assessment has been carried out on the audit publication to identify risks to patient confidentiality. Rounding has been used for numbers derived from harms records to reduce the risk of patient identification; numbers between 1 and 7 (inclusive) are shown as 5, while all other numbers are rounded to the nearest 5. Percentages will be calculated using rounded numerators and denominators. The calculated value will not be shown in cases where the denominator is less than 20. Statistical calculations (e.g. significance tests) are performed on unrounded numbers.

In line with the government transparency agenda, NHS Digital provides local participation data taken from the audit in CSV format as part of the audit publication available through the NHS Digital website, and also through the UK open data portal, https://data.gov.uk.

The main report is published in web-based format (html). Accompanying versions of the report in both PowerPoint and PDF formats are published as resources on the summary page. However, future NDISA publications will be web-based only. This is to improve the accessibility of our publications.

A CSV file of the backing data used in the report is available on the NHS Digital website and through www.data.gov.uk. The data is aggregated at national level (England or England and Wales).

The NDISA 2018-21 report. 

Further information about the NDISA harms collection.

Information about the superseded NaDIA harms collection. 

The final NaDIA harms audit report from 2021.

Information about the closed NaDIA snapshot.

The final NaDIA snapshot report from 2019. 

The NDISA advisory group (consisting of patient representatives, healthcare professionals, administrators, researchers and analysts, including representation from Diabetes UK and NHS Digital) provide advice on the content of the reports as well as the direction and development of the audit.

The wider NDA team has an active role in the National Cardiovascular Intelligence Network (NCVIN) workshops to gain a better understanding of how commissioners and localities use the data and how we can improve the NDA programme’s publications and supporting information.

The NDISA harms collection is designed to be a low-burden collection, with only 4 data items required from submitters:

• NHS number: for data validation and linkage
• Harm type: hypoglycaemia requiring rescue treatment, diabetic ketoacidosis (DKA), hyperglycaemic hyperosmolar state (HHS), new diabetic foot ulceration (DFU)
• Date the harm occurred
• Hospital site at which the harm occurred

The audit team acknowledges that participation in the audit involves costs in both time and organisation for the providers that take part, and thanks them for their efforts. The audit continues to look at ways in which to reduce respondent burden and increase ease of participation and welcomes comments and suggestions.

NHS Digital is keen to gain a better understanding of the users of this publication and their needs, as also welcomes suggestions on how to decrease respondent burden.

Your feedback is welcome and may be sent to [email protected]. Please include ‘National Diabetes Inpatient Safety Audit’ in the subject line.

Alternatively, you can call our contact centre on 0300 303 5678

Or write to:

NHS Digital,
7-8 Wellington Place,
Whitehall Road,
Leeds
LS1 4EG.