Publication, Part of National Diabetes Inpatient Safety Audit (NDISA)
National Diabetes Inpatient Safety Audit (NDISA) 2018-2021
Audit
Inpatient harms: Patient profiles
Background
This section of the report looks at characteristics associated with a greater risk of the 4 inpatient harms collected by the National Diabetes Inpatient Safety Audit (NDISA), with the aim of better understanding high-risk features which might help target preventive care.
The hospital admission characteristics of people with diabetes are identified in the Hospital Episode Statistics (HES) dataset. Patient demographics, diabetes characteristics, treatment targets and care processes are linked from the core National Diabetes Audit (NDA).
Data from the NDISA harms collection is used to identify hospital bed days where 1 of the following inpatient harms occurred: hypoglycaemic rescue, diabetic ketoacidosis (DKA), hyperosmolar hyperglycaemic state (HHS) and diabetic foot ulcer (DFU). A baseline population of inpatients with diabetes is created for comparison, excluding bed days where an inpatient harm occurred.
To adjust for length of stay, figures are based on occupied bed days. For example, a hospital stay lasting 1 night counts as 1 occupied bed day, 5 nights counts as 5 occupied bed days etc. The statistical significance of differences between occupied bed days where a harm occurred versus the bulk of days where a harm did not occur was assessed at the 95% level.
- Have been admitted as an emergency (see Table 5.4).
- Have type 1 diabetes (Table 5.2).
- Have not received 8 care processes in the last 12 months (Table 5.3).
- Have not met the HbA1c treatment target (Table 5.3).
- Experience cardiovascular or diabetes-specific complications on admission or during their hospital stay (Table 5.5). In particular, inpatients having strokes on or during admission are more likely to experience HHS during their hospital stay.
Summary
Table 5.1 (below) compares inpatient characteristics where a harm occurred against the characteristics of the whole inpatient population with diabetes. For example, those experiencing inpatient DKA are younger than the wider inpatient population with diabetes.
If inpatient harms were compared against the whole inpatient population (including those without diabetes) or against the population as a whole, different results would be found. For example, it is almost certain that each harm would be associated with higher deprivation.
Table 5.1: Summary of characteristics associated with each inpatient harm compared to the whole inpatient population with diabetes1, England, May 2018 - October 2021
| Characteristic | Inpatient harm | ||||
| Hypoglycaemic rescue | DKA | HHS | DFU | ||
| Demographics | Age | Younger | Younger | Older | None |
| Sex | None | Female | Male | Male | |
| Ethnicity | White | White | None | White | |
| Smoking status2 | Current | Current | None | None | |
| Deprivation quintile | Least | Least | None | None | |
| BMI | Lower | Lower | Lower | None | |
| Diabetes characteristics | Diabetes type | Type 1 | Type 1 | Type 1 | Type 1 |
| Diabetes duration | Longer | Longer | Longer | Longer | |
| Renal function | Worse | Better | None | Worse | |
| Treatment targets and care processes3 | Blood pressure (≤ 140/80) | None | None | None | None |
| Cholesterol (< 5 mmol/L) | Not met | Not met | None | None | |
| HbA1c (≤ 58 mmol/mol) | Not met | Not met | Not met | Not met | |
| Met 3 treatment targets? | No | No | No | No | |
| Had all 8 care processes? | No | No | None | No | |
| Hospital admissions | Admission method4 | Emergency | Emergency | Emergency | Emergency |
| Admission type | Medical | Surgical | Medical | Surgical | |
| Complications during admission5 | Heart failure, Admitted for DKA, Admitted with diabetic foot disease, Renal replacement therapy (RRT) required during admission | Angina, Heart failure, Admitted for DKA, RRT required during admission | Stroke, Admitted for DKA, Admitted with diabetic foot disease | Heart failure, Admitted for DKA, Admitted with diabetic foot disease, RRT required during admission | |
Notes:
1. Statistical significance of harms population vs. inpatient population with diabetes at the 95% level. Proportions are tested using the Chi-squared test. Medians are tested using the Mann–Whitney U test.
2. Group comprises of non-smokers whose history is unknown and patients who have never smoked.
3. Treatment target and care process information was taken from the core NDA audit year prior to the hospital admission (e.g. from 2017-18 where the person’s hospital admission was in 2018-19).
4. Admission method = The method of admission to hospital e.g. emergency or elective. An elective admission is one that has been arranged in advance. Admission type is defined by the specialty under which consultant is contracted e.g. surgical or medical. The category ‘Other’ covers: Admission methods ‘Maternity’ and ‘Other’; Admission types ‘Other’, ‘Pathology’, ‘Psychiatry’ and ‘Radiology’.
5. Complication recorded at any point during the admission, except: ‘Admitted for DKA’ and ‘Admitted with diabetic foot disease’. The definition of Admitted for DKA only includes admissions where DKA was the 1st diagnosis code on the 1st episode of the admission. The 2019 NaDIA harms audit report included cases where DKA was in any diagnosis position on the 1st episode of the hospital admission.
Diabetes characteristics
Table 5.2: Diabetes characteristics, by inpatient harm2, England, May 2018 - October 2021 (rounded1)
| Group | Diabetes Type | Diabetes Duration | Renal function (Estimated glomerular filtration rate - eGFR) (ml/min/1.73m2) | ||||
| Type 1 | Type 2 | Median | Median | ||||
| % | % | ||||||
| Inpatient population with diabetes | 7.7 | 92.3 | 12 | 67.7 | |||
| Hypoglycaemic rescue | 34.1 | 65.9 | * | 19 | * | 58.8 | * |
| DKA | 63.5 | 36.5 | * | 21 | * | 77.2 | * |
| HHS | 13.9 | 86.1 | * | 16 | * | 66.8 | n |
| Inpatient population with diabetes3 (Length of stay - LOS≥3) |
7.5 | 92.5 | 12 | 67.2 | |||
| DFU (LOS≥3) | 14.3 | 86.6 | * | 16 | * | 58.2 | * |
- Diabetes type 1
- Longer diabetes duration
- Impaired renal function (except DKA and HHS)
Notes:
* = statistically significant at the 0.05 level vs. inpatient population. n = not statistically significant. Proportions are tested using the Chi-squared test. Medians are tested using the Mann–Whitney U test. Cases with missing or unknown values are excluded from the calculations. The proportions of the inpatient population (data row 1, all diabetes) with missing or unknown values are: Diabetes type 3.5%; Diabetes duration 1.9%; eGFR 13.2%.
1. Percentages are derived from rounded values. Underlying counts between 1 and 7 are set to 5. All counts greater than 7 are rounded to the nearest 5. Consequently some percentages may not sum up to exactly 100%.
2. Proportions are calculated from the sum of nights in hospital during the period stated for people in the core NDA, where diabetes was diagnosed on or before admission. Day cases and same-day discharges are counted as zero days and are therefore excluded. For further information, see: Further information: Inpatient population with diabetes.
3. See note 2 above, with additional exclusion for admissions that are less than 3 nights due to the audit requirement that new onset foot ulcers must occur more than 72 hours after admission.
Treatment targets and care processes
Table 5.3: Treatment targets and care processes in the audit year preceding hospital admission4, by inpatient harm2, England, May 2018 - October 2021 (rounded1)
| Group | Treatment targets4 in year prior to admission | Received all 8 care processes4 in year prior to admission | ||||||||
| Blood pressure | Cholesterol | HbA1c | Met all 3 treatment targets | |||||||
| (≤ 140/80) | (< 5 mmol/L) | (≤ 58 mmol/mol) | ||||||||
| % | % | % | % | % | ||||||
| Inpatient population with diabetes | 67.6 | 67.7 | 56.3 | 31.4 | 43.5 | |||||
| Hypoglycaemic rescue | 68.1 | n | 64.2 | * | 26.0 | * | 14.6 | * | 37.8 | * |
| DKA | 67.0 | n | 56.3 | * | 14.2 | * | 7.1 | * | 37.6 | * |
| HHS | 68.8 | n | 68.8 | n | 34.4 | * | 18.8 | * | 40.6 | n |
| Inpatient population with diabetes3 (LOS≥3) | 67.5 | 67.7 | 56.3 | 31.4 | 43.0 | |||||
| DFU (LOS≥3) | 65.5 | n | 67.3 | n | 44.2 | * | 22.1 | * | 36.3 | * |
- Higher HbA1c in the audit year prior to admission.
- Not having all 8 care processes in the audit year prior to admission (except HHS).
The low proportion for each harm not meeting all 3 treatment targets is primarily driven by those missing the HbA1c target.
Notes:
* = statistically significant at the 0.05 level vs. inpatient population. n = not statistically significant. Proportions are tested using the Chi-squared test. Cases with missing or unknown values are excluded from the calculations. The proportions of the inpatient population (data row 1, all diabetes) with missing or unknown values are: 8.6-8.7% (all variables).
1. Percentages are derived from rounded values. Underlying counts between 1 and 7 are set to 5. All counts greater than 7 are rounded to the nearest 5. Consequently some percentages may not sum up to exactly 100%.
2. Proportions are calculated from the sum of nights in hospital during the period stated for people in the core NDA, where diabetes was diagnosed on or before admission. Day cases and same-day discharges are counted as zero days and are therefore excluded. For further information, see: Further information: Inpatient population with diabetes.
3. See note 2 above, with additional exclusion for admissions that are less than 3 nights due to the audit requirement that new onset foot ulcers must occur more than 72 hours after admission.
4. Treatment target and care process information was taken from the core NDA audit year prior to the hospital admission (e.g. from 2017-18 where the person’s hospital admission was in 2018-19).
Hospital admissions
Table 5.4: Hospital admission method and main specialty, by inpatient harm2, England, May 2018 - October 2021 (rounded1)
| Group | Admission method4 | Admission type4 | ||||||
| Emergency | Elective | Other | Medical | Surgical | Other | |||
| % | % | % | % | % | % | |||
| Inpatient population with diabetes | 83.9 | 9.8 | 6.3 | 71.7 | 22.1 | 6.3 | ||
| Hypoglycaemic rescue | 93.8 | 3.2 | 3.1 | * | 81.3 | 18.3 | 0.4 | * |
| DKA | 91.4 | 6.1 | 2.5 | * | 69.2 | 30.3 | 0.5 | * |
| HHS | 91.4 | 2.9 | 2.9 | * | 85.7 | 14.3 | 0 | * |
| Inpatient population with diabetes3 (LOS≥3) | 84.1 | 9.1 | 6.9 | 71.8 | 20.9 | 7.4 | ||
| DFU (LOS≥3) | 89.2 | 5.8 | 5 | * | 70 | 29.2 | 0.8 | * |
- Higher risk characteristics for inpatient harms include emergency admission.
- Hypoglycaemic rescue and HHS are associated with medical admission and DKA and DFU are associated with surgical admissions.
Notes:
* = statistically significant at the 0.05 level vs. inpatient population. n = not statistically significant. Proportions are tested using the Chi-squared test. Cases with missing or unknown values are excluded from the calculations. The proportions of the inpatient population (data row 1, all diabetes) with missing or unknown values are: Admission method 0.0%; Admission type 0.2%.
1. Percentages are derived from rounded values. Underlying counts between 1 and 7 are set to 5. All counts greater than 7 are rounded to the nearest 5. Consequently some percentages may not sum up to exactly 100%.
2. Proportions and rates are calculated from the sum of nights in hospital during the period stated for people in the core NDA, where diabetes was diagnosed on or before admission. Day cases and same-day discharges are counted as zero days and are therefore excluded. For further information, see: Further information: Inpatient population with diabetes.
3. See note 2 above, with additional exclusion for admissions that are less than 3 nights due to the audit requirement that new onset foot ulcers must occur more than 72 hours after admission.
4. Admission method = The method of admission to hospital e.g. emergency or elective. An elective admission is one that has been arranged in advance. Admission type is defined by the specialty under which consultant is contracted e.g. surgical or medical. The category ‘Other’ covers: Admission methods ‘Maternity’ and ‘Other’; Admission types ‘Other’, ‘Pathology’, ‘Psychiatry’ and ‘Radiology’.
Complications
Table 5.5: Complications during hospital admission, by inpatient harm2, England, May 2018 - October 2021 (rounded1)
| Group | Cardiovascular complications4 | Diabetes-specific complications4 | ||||||||||||
| (on or during admission) | (at specified point during admission) | |||||||||||||
| Angina | Myocardial infarction | Heart failure | Stroke | Admitted for DKA (5,6) | Admitted with diabetic foot disease | RRT required during admission | ||||||||
| % | % | % | % | % | % | % | ||||||||
| Inpatient population with diabetes | 7.9 | 3.6 | 21.8 | 7.5 | 1 | 5.1 | 4.9 | |||||||
| Hypoglycaemic rescue | 7.3 | n | 4.1 | n | 26.6 | * | 7.5 | n | 5.6 | * | 7.5 | * | 8.5 | * |
| DKA | 5.1 | * | 4.5 | n | 12.6 | * | 7.6 | n | 22.7 | * | 5.1 | n | 8.1 | * |
| HHS | 5.7 | n | 5.7 | n | 22.9 | n | 28.6 | * | 2.9 | * | 2.9 | * | 5.7 | n |
| Inpatient population with diabetes3 (LOS≥3) | 7.6 | 3.6 | 22.6 | 8.2 | 0.9 | 5.4 | 5.2 | |||||||
| DFU (LOS≥3) | 6.7 | n | 4.2 | n | 33.3 | * | 9.2 | n | 1.7 | * | 15.8 | * | 10 | * |
- Most cardiovascular or diabetes-specific complications are associated with inpatient harms.
- In particular, inpatients having strokes on or during admission are more likely to experience HHS during their hospital stay.
Notes:
* = statistically significant at the 0.05 level (vs. inpatient population with diabetes). n = not statistically significant (vs. Inpatient population with diabetes). Proportions are tested using the Chi-squared test.
1. Percentages are derived from rounded values. Underlying counts between 1 and 7 are set to 5. All counts greater than 7 are rounded to the nearest 5. Consequently some percentages may not sum up to exactly 100%.
2. Proportions and rates are calculated from the sum of nights in hospital during the period stated for people in the core NDA, where diabetes was diagnosed on or before admission. Day cases and same-day discharges are counted as zero days and are therefore excluded. For further information, see: Further information: Inpatient population with diabetes.
3. See note 2 above, with additional exclusion for admissions that are less than 3 nights due to the audit requirement that new onset foot ulcers must occur more than 72 hours after admission.
4. Complication recorded at any point during the admission, except: ‘Admitted for DKA’ and ‘Admitted with diabetic foot disease’.
5. The definition of Admitted for DKA only includes admissions where DKA was the 1st diagnosis code on the 1st episode of the admission. The 2019 NaDIA harms audit report included cases where DKA was in any diagnosis position on the 1st episode of the hospital admission.
Last edited: 14 July 2022 9:33 am