The cost of diabetes-related hospital care to the Southern District Health Board in 2016/17

Diabetes is an increasingly common non-communicable disease associated with high personal and health system costs. The global prevalence of diagnosed diabetes was estimated to be 8.4% among adults aged 18 years and over in 2017, and this is projected to increase to 9.9% by 2045.1 In New Zealand, the prevalence of diabetes among those aged 15 years and over was 7.0% in 2008/09,2 and this is likely to have increased over the last 10 years alongside the increasing prevalence of obesity.3 In 2017, there were an estimated 245,680 New Zealanders with diagnosed diabetes.4

An estimated 12% of global health expenditure is spent on diabetes.5 In 2017 this cost was more than US$850 billion.6 Health system costs of diabetes include the management of diabetes itself, as well as the associated complications such as cardiovascular disease, diabetic retinopathy, renal failure and peripheral vascular disease. Compared to individuals without diabetes, those with diabetes are more likely to be hospitalised for any reason,7 and a large portion of health system costs for diabetes is attributable to hospital care.8

In New Zealand, while direct pharmaceutical costs for diabetes are known and monitored annually ($79.7 million in 2018),9 there are few published data on the other health system costs associated with diabetes. A now decade-old report by PricewaterhouseCoopers (PWC), first produced in 2001 and updated in 2007, estimated diabetes cost NZ$540 million (NZ$1=US$0.66 as at 10 June 2019) in 2007, and forecast that diabetes would cost between NZ$1–1.2 billion by 2016/17.10,11 The cost to the Canterbury District Health Board (DHB) for hospital admissions in 2007, where diabetes was recorded as the primary or secondary diagnosis, was estimated at NZ$10.1 million.12 At Counties Manukau DHB, cardiovascular disease, diabetes or both, were the primary reasons for 13% of hospitalisations for those aged 15 years and over, yet together contributed to 46% of the total inpatient hospitalisation cost of NZ$101 million in 2008.13 This cost increased to NZ$151 million, if pharmaceuticals and laboratory services were also included, of which NZ$83 million was for those with diabetes. For the period 2007–2014, an estimated 5.3% of total health expenditure in New Zealand was for type 2 diabetes alone.14

The aim of this study was to estimate the cost to the Southern DHB of diabetes-related inpatient admissions to Dunedin Hospital and Southland Hospital for the 2016/17 financial year (1 July 2016–30 June 2017).

Methods

Data for all admissions to Dunedin Hospital and Southland Hospital with any International Classification of Diseases (ICD) diabetes diagnostic code were obtained from the Southern DHB for the financial year ending June 2017. For the study period, all diabetes-related inpatient hospital admissions were coded using the ICD 10th Revision Australian Modification (ICD-10-AM) codes.15 The ICD-10-AM diabetes codes used in this study were E10 Type 1 diabetes, E11 Type 2 diabetes, E13 Other specified diabetes, E14 Unspecified diabetes, O240-244 Pre-existing diabetes in pregnancy and O249 Diabetes diagnosed in pregnancy. As well, data for all admissions with a primary diagnosis of coronary artery disease (ICD-10-AM codes I20-I25) during the same 12-month period, irrespective of diabetes status, were obtained.

A unique non-identifying number was assigned to each admission. The data accessed for each admission included: admission and discharge dates, date of birth, sex, ethnicity, hospital (Dunedin or Southland), all ICD-10-AM codes for all diagnoses associated with each admission, the order of the diagnoses according to the primary, secondary or subsequent reason(s) for admission, the Diagnosis-Related Group (DRG) codes, the primary procedure and associated code (where applicable) and the case weight. Case weights measure the relative complexity of the treatment(s) given to patients during their hospital admission.16 Differences in case weights reflect the resources needed for each admission, such as number of days in hospital or the length of time in the operating theatre. For example, in this study the case weight for a coronary bypass with invasive cardiac investigation and reoperation was 12.3 compared with 0.22 for congestive heart failure.

Data analysis

Data were for separate admission events. The diabetes admission data were categorised into three categories according to whether the diabetes diagnostic code was listed first (primary diagnosis), second (secondary diagnosis) or subsequently (‘other’ diagnosis). A primary diagnosis was considered to be the main reason for the hospital admission and a secondary diagnosis was the main underlying reason for the hospital admission. Other diagnoses were any other listed conditions, but these were not necessarily in any order of priority. Therefore, for example, where a single admission event had nine ICD codes listed and diabetes was neither the primary or secondary diagnosis, whether the diabetes code was listed fifth or seventh or eighth, the order had no bearing on the relative importance of diabetes in relation to that hospital admission.

Where diabetes was coded as the primary reason for hospital admission, these admissions were further categorised according to the type of complication, such as renal complications or neurological complications. Admissions where diabetes was the secondary or an ‘other’ diagnosis were categorised according to the primary diagnosis ICD-10-AM code into the main disease groups, for example, C00-D48 Neoplasms, I00-I99 Diseases of the circulatory system. The coronary artery disease admission data were categorised into those with a diabetes diagnostic code listed, and those without.

The cost of each admission was calculated by multiplying the case weight by the 2016/17 cost weight value of NZ$4,824.67. Separate admission costs were summed for each of the different diagnostic categories, as described above.

As the demographic data related to admissions and not individuals, matching data based on birthdate, sex, ethnicity and type of diabetes was undertaken to estimate the number of individuals who were hospitalised during the study period.

Māori consultation was undertaken with both the Ngāi Tahu Research Consultation Committee, University of Otago and the Southern DHB before commencing this study. Ethics approval was obtained the Human Research Ethics Committee of the University of Otago (HD17/066). Data were analysed in Excel (Microsoft) and Stata SE 15 (StataCorp).

Results

The total number of diabetes-related admissions to Dunedin and Southland Hospitals for the year ended 30 June 2017 was 6,994. These admissions were for an estimated 3,615 individuals with diabetes. This means that some individuals had multiple admissions and were included more than once in the demographic data relating to admissions. The median number of diagnoses per admission was six, with a range of 1–73.

In Table 1 the demographic characteristics are presented by admission event. The demographic characteristics of the estimated 3,615 individuals are shown in Appendix Table 1. Among these individuals, 59.4% had one admission, 20.7% had two admissions and 19.9% had three or more admissions, with one having 31 admissions. Of the 6,994 admissions, nearly 65% were in Dunedin Hospital, and 80% were for patients aged 55 years and older. Diabetes was the primary diagnosis for only 7% of admissions, of which 31% were associated with type 1 diabetes. In contrast, diabetes was listed as an ‘other’ diagnosis for 58% of admissions, and nearly 86% of all admissions were associated with type 2 diabetes.

Table 1: Demographic characteristics and cost of admissions to Dunedin and Southland hospitals for the 2016/17 financial year by primary, secondary and ‘other’ diabetes diagnosis.

‍

The estimated total cost of all diabetes-related admissions was NZ$40,968,618, with 75% of this cost associated with admissions where diabetes was listed as an ‘other’ diagnosis (Table 1). As summarised in Table 2, the total cost of admissions relating to diabetes as a primary diagnosis was NZ$2.2 million, of which 68% (NZ$1.5 million) was associated with type 2 diabetes. Ketoacidosis was the most frequent reason for admission (n=103) among those with type 1 diabetes at a cost of NZ$349,892. Where type 2 diabetes was the primary diagnosis, admissions with the highest cost were attributable to peripheral circulatory complications (NZ$394,944) and type 2 diabetes with multiple complications (NZ$698,062).

Table 2: Cost of admissions to Dunedin and Southland hospitals where diabetes was the primary diagnosis sorted by ICD-10-AM† diabetes code for the year to 30 June 2017.

‍

†International Classification of Diseases 10th Revision Australian Modification (ICD-10-AM).

As detailed in Table 3, the total cost of admissions where diabetes was the secondary diagnosis was slightly over NZ$8 million. The disease categories with the highest cost were ‘diseases of the eye and adnexa’ (NZ$1.2 million), followed by ‘diseases of the circulatory system’ (NZ$1.16 million), ‘diseases of the musculoskeletal system and connective tissue’ (NZ$992,883) and ‘neoplasms’ (NZ$655,339). Most of the total cost (NZ$30.7 million) of all diabetes-related admissions was for admissions where diabetes was not coded as a primary or secondary diagnosis. Of this, NZ$8.2 million was primarily for diseases of the circulatory system, and NZ$4.5 million for reasons coded as ‘injury, poisoning and certain other consequences of external causes’, with more than NZ$2 million for each of the following categories: ‘factors influencing health status and contact with health services’, ‘diseases of the digestive system’, ‘diseases of the respiratory system’ and ‘neoplasms’.

Table 3: Costs of admissions to Dunedin and Southland hospitals where diabetes was a secondary or ‘other’ diagnosis categorised by ICD-10-AM† primary diagnostic category for the year to 30 June 2017.

‍

†International Classification of Diseases 10th Revision Australian Modification (ICD-10-AM).

There were 1,573 admissions with a primary diagnosis of coronary artery disease, of which 371 (24.9%) also had a diabetes diagnostic code. The mean (SD) length of stay for those with a diabetes code was 4.1 (8.3) days compared with 3.0 (5.0) days for those without a diabetes code. Similarly, the mean (SD) cost per admission with and without diabetes codes was NZ$10,407 ($20,694) and NZ$8,657 ($11,347), respectively.

Discussion

The prevalence of diabetes and the costs of treating this disease continue to rise both internationally and in New Zealand.6 The number of people with diabetes on the New Zealand Virtual Diabetes Register was 187,860 in 2010 increasing to 245,680 in 2017, and from 12,002 to 14,355 in the Southern DHB area over the same period.4 The annual cost of diabetes medicines and diabetes management (which includes blood glucose monitoring metres and testing strips) are monitored by PHARMAC and the cost of these treatments was NZ$74.5 million for the year to 30 June 2017,17 increasing by 7% to NZ$79.9 million for the year to 30 June 2018.9 In contrast, diabetes hospital and primary care costs are not regularly monitored in New Zealand. We sought to estimate the cost of diabetes-related hospital admissions to Dunedin Hospital and Southland Hospital for the 2016/17 financial year. This cost was NZ$41 million (for a resident population of about 320,640),18 of which NZ$2.2 million was for admissions where diabetes was the primary diagnosis, and NZ$8 million where diabetes was the secondary diagnosis. These costs are most likely to be an underestimate, as it is recognised that diabetes is under-reported on hospital admission data in New Zealand and overseas.19–21 The degree of under-reporting in this study is not known.

There are few published studies in New Zealand with which to compare our results. Sheerin estimated the cost to the Canterbury DHB of hospitalisations where diabetes was the primary or secondary diagnosis.12 This cost was NZ$10.1 million in 2005/06, which is the same cost associated with primary and secondary diabetes diagnoses estimated in our study. However, in addition to the two studies being over 10 years apart, there are other differences which limit direct comparisons. The population for the Canterbury region is about twice that for the Otago/Southland area. In 2006, the population for Canterbury was 521,832 and the combined population for the Otago/Southland region was 284,673.22 There are also methodological differences. First, our study included hospitalisations for pre-existing or newly-diagnosed diabetes in pregnancy, unlike the Canterbury study.12 Second, we included all disease categories where diabetes was the secondary diagnosis, whereas the Canterbury study only included those codes which were considered to be directly related to known complications of diabetes-acidosis (E87.2), diseases of the nervous system (G45-63), diseases of the eye and adnexa (H25-41), diseases of the circulatory system (I20-75), diseases of the genitourinary system (N10-23), and preparatory care for dialysis (Z49.0).12 Primary reasons for admission such as osteomyelitis where diabetes is an underlying contributing factor or lower limb amputations for instance, were not included in the Canterbury study, resulting in an underestimate of costs.

As illustrated in our study and the Canterbury study,12 diabetes is typically not considered the primary reason for hospitalisation, yet is frequently coded as the secondary diagnosis or a co-morbid condition. This was particularly evident among those aged 65 years and over, for whom diabetes was recorded as a secondary (31.4%) or ‘other’ (62.3%) diagnosis for most admissions. These admissions for this age group cost 65% of the total NZ$41 million cost. This is almost the same as the US, where 64% of hospital inpatient expenditure is for those aged 65 years and over.23

Health system costs of individuals with diabetes are greater than that of those without diabetes, irrespective of whether diabetes is the primary, secondary or subsequent diagnosis. In Australia, the annual cost (including medications, hospitalisations and ambulatory services) of a patient with diabetes is 2.3 times more than a patient with normal glucose tolerance (A$4,390 compared with A$1,898; (NZ$1=A$0.95 as at 10 June 2019).24 In New Zealand, while data are limited, the total additional cost to the Counties Manakau DHB for those with diabetes compared with those without diabetes who were hospitalised during 2007 was NZ$66 million.13 A significant part of the additional cost for patients with diabetes is due to longer hospital stays,25 as we demonstrated in this study for admissions where coronary artery disease was the primary diagnosis. Similarly, a case control study examining lower limb cellulitis risk factors conducted at Auckland City Hospital found type 2 diabetes was associated with a significantly longer hospital stay compared with those without diabetes (median 5.3 vs 3.0 days, P<0.001) regardless of age and ethnicity.26 Moreover, in this Auckland study, those with type 2 diabetes (20% of the identified cases) were more likely to be re-admitted further increasing hospital costs. The average annual cost of lower limb cellulitis was estimated to be A$4.2 million of which A$1.4 million was for type 2 diabetes patients.

Clinical coding practice is governed by rules and conventions to ensure consistency and accuracy of information. These rules are updated over time. In the eighth revision of the ICD-10-AM, it became a requirement to code diabetes whenever a patient with diabetes is hospitalised to recognise that on average they require a higher (more expensive) standard of care.15 However, unless diabetes is a primary or secondary diagnosis, it is not prioritised in the list of subsequent codes. As such, it is not possible to determine the extent to which diabetes contributes to the reason for admission. In our study, how much diabetes contributed to the NZ$30.7 million cost where diabetes was listed as an ‘other’ diagnosis was not able to be determined, but it does, however, highlight that hospital costs for those with diabetes are substantial.

Diabetes is a risk factor for many diseases, and while it may be the underlying or a contributing cause, this is not always reflected in hospital admission or mortality data. Indeed, diabetes may not be recorded at all.19,25,27 In 2007 in Scotland, only 59% of hospital admissions for people known to have diabetes prior to admission had a diabetes code recorded.19 Because of this under-recording, the health system cost of diabetes can be underestimated. In our study, a disease of the circulatory system (including ischaemic heart disease, stroke and peripheral vascular disease) was the most common primary diagnosis, and the cost of these hospital admissions was NZ$9,342,777 or 23% of the total diabetes-related admission cost. For most of these admissions diabetes was not recorded as either the primary or secondary diagnosis. Indeed, diabetes was recorded as a secondary diagnosis in only 12% of admissions where a circulatory disease was coded as the primary diagnosis, yet diabetes is often the underlying cause of many cardiovascular disorders.

A limitation to the scope of the study was the inclusion of hospital inpatient costs only. Including the cost of out-patient appointments, retinal screening, home renal dialysis, private hospital admissions, pharmaceuticals prescribed outside the index admission and primary care consultations would provide a more complete description of the health system cost of diabetes. Also, data from the smaller hospitals in the Southern DHB region (Oamaru, Balclutha, Clutha and Queenstown) were not available at the time of sourcing the main dataset, although the number of diabetes-related admissions at these hospitals is likely to be relatively small.

This study estimated that the cost of diabetes in 2016/17 to the Southern DHB amounted to NZ$10 million when diabetes was classified as a primary or secondary diagnosis. However, the actual cost of diabetes to the Southern DHB far outweighs this value. Attributing costs related to diseases of the circulatory system to diabetes, and other diabetes-related co-morbidities remains challenging. Given the continued increase in the prevalence of diabetes in New Zealand, monitoring the cost of diabetes to DHBs should be prioritised, along with the implementation of interventions that target preventable diabetes-related hospital admissions, and diabetes prevention intervention programmes.

Appendix Table 1: The demographics and diabetes type for the total number of admissions to Dunedin and Southland hospitals and the estimated number of individuals for the 2016/17 financial year.