Design and setting

This retrospective observational study uses data from all individuals who died in 2012 in Belgium. Supportive measures for palliative home care exist in Belgium since 1985, and have since been further expanded [15]. Since 2002, palliative care is legally recognised as a right in Belgium.

Data

Cohort data from eight routinely collected population-level databases, handled by three different organisations, were linked. The data consists of (1) the socio-demographic database of all individuals with healthcare insurance (legally mandatory in Belgium); (2) the health care database containing all reimbursed health care use data on home, nursing home, outpatient and hospital care; (3) the pharmaceutical database containing all reimbursed medication data; (4) Belgian Cancer Registry data with diagnostic information on all incidences of cancer including date of diagnosis and type of cancer; (5) death certificate data containing cause of death; (6) population registry data including nationality and household composition; (7) census data, including educational level and housing characteristics and (8) the fiscal database (including net taxable income).

After acquiring approvals from the relevant data protection agencies, the databases were linked for analysis in a secure and ethically responsible manner, guaranteeing the anonymity of the deceased. This process is described elsewhere [16].

Subjects

To limit the analysis to those who can theoretically use the supportive measures, we excluded minors and people residing in a nursing home before death from the dataset.

To further limit the analysis to a population that is most likely to benefit from the measures, we made a second subpopulation of people who died from illnesses highly indicative of palliative care needs, defined by Rosenwax et al[14] as the ‘Minimal Estimate of potential users of palliative care’ through mixed-methods research, referred to here as the palliative subset. The following underlying causes of death were selected using the 10th revision of the International Statistical Classification of Diseases and Related Health Problems (ICD): neoplasms (ICD-10 C00-D48), chronic obstructive pulmonary disease (ICD-10 J40-44, J47), other organ failures (i.e. heart, renal, and liver failure) (ICD-10 I11-I13, I50, K70-72, N10-12, N18-19), neurodegenerative diseases (i.e. Alzheimer’s, Parkinson’s, motor neurone, and Huntington’s disease) (ICD-10 F01, F03, G10, G12, G20, G30), and HIV/aids (ICD-10 B20-24).

Terms and classification of supportive measures

We defined supportive measures for palliative home care as: ‘all health care allowances and services (in addition to the standard primary care services) that can support the patient to remain at home in the last phase of life’. We divided supportive measures for palliative home care into statutory and non-statutory:

Sociodemographic characteristics

Population characteristics available through the datasets are: age at death, sex, nationality, household type, housing comfort, educational level (using UNESCO’s International Standard Classification of Education (ISCED)), region, degree of urbanisation, annual personal net taxable income and underlying cause of death (using the 10^th revision of the International Classification of Diseases (ICD-10), categorized as: neoplasms, COPD, other organ failure, neurodegenerative disease, HIV/aids, other). For codification clarifications and detailed information on data administrators and data sources see S1 Appendix.

Data analysis

Descriptive statistics were used to describe population characteristics in the full population and in the palliative subset. To compare the chances of receiving a policy measure to support palliative home care between different population groups a multivariable binary logistic regression was performed with uptake (of a statutory measure, a non-statutory measure, or one or more of all measures) as the dependent variable. A hierarchical approach was followed to build the model (see S1 Appendix). An alternative model was analysed on a subgroup of only cancer deaths (see S1 Table). The analyses were generated using the SAS Enterprise Guide 7.1 software.

Ethics approvals

This study received approvals by the Commission for Medical Ethics of the University Hospital Brussels (B.U.N. 143201629410). Trusted Third Parties (TTPs) ‘eHealth’ and ‘Crossroads Bank for Social Security (CBSS)’ were responsible for the deidentification process and deterministic one-to-one record linkage of the databases. In accordance with Belgian law, approvals for access to and integration of the various databases were obtained from two separate national sectoral committees for privacy protection: the ‘Sectoral Committee of Social Security and Health, Section Health’ and the ‘Statistical Supervisory Committee’. Both are subcommittees of the Belgian Commission for the Protection of Privacy. No informed consent was required. For a full description of the data linkage procedure, including details on the data anonymization, see Maetens et al.[16].

Population characteristics

In 2012, 87,007 health-insured adult persons died in Belgium while living at home (Table 1). Of these, 38,657 died from a cause indicative of palliative care need: 29.5% died from a neoplasm, 4.3% from COPD, 5.6% from another organ failure, 5% from neurodegenerative disease, and 0.04% from HIV/aids.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of all deaths and deaths of persons with potential palliative care needs (in %).

https://doi.org/10.1371/journal.pone.0213731.t001

Uptake of supportive measures for palliative home care

Of all home-dwelling adults, 17.9% used a statutory measure, 51.5% a non-statutory measure, 55.2% at least one supportive measure, and 14.1% at least one statutory and non-statutory measure (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. The use of supportive measures for palliative home care (in %).

https://doi.org/10.1371/journal.pone.0213731.t002

Of those in the palliative subset, 33.1% used a statutory and 62.2% a non-statutory measure; 69.7% used at least one of each, and 25.7% used both (see S2 Table). The statutory and non-statutory measures were used by 42.7% and 62.3% respectively of those dying of neoplasms, 11.4% and 65.6% of COPD, 13.1% and 61.7% of other organ failures, 17.7% and 59.8% of neurodegenerative diseases, and 5.9% and 32.4% of HIV/aids.

Characteristics associated with uptake of a supportive measure for palliative home care

Multivariable logistic analysis for the palliative subset indicated that those under 65 years were significantly more likely to use a statutory measure compared to those above 95 years old (Odds Ratio = 1.32; 95% Confidence Interval [1.07–1.62]), but not a non-statutory measure (Table 3). The age groups between 65 years old and 94 years old were not significantly more or less likely to use a statutory measure compared to those aged over 95, but significantly more likely to use a non-statutory measure. Age was not a significant predictor for the use of either a statutory or non-statutory measure (third model). Women were more likely to use a statutory (OR = 1.10[1.04–1.16]) or non-statutory (OR = 1.35[1.28–1.41]) measure than men. Compared with those who died from neoplasms, people who died from COPD (OR = 0.18[0.16–0.20]), other organ failures (OR = 0.22[0.20–0.24]), neurodegenerative disease (OR = 0.30[0.27–0.33]), or HIV/aids (OR = 0.17[0.04–0.72]) were less likely to use a statutory measure. Compared with people who died from neoplasms, those who died from COPD were more likely to use a non-statutory measure (OR = 1.22[1.13–1.32]), but those who died from neurodegenerative diseases (OR = 0.86[0.80–0.92]) or from HIV/aids (OR = 0.41[0.18–0.94]) were less likely to receive such a measure. Results predicting the use of a non-statutory measure were not significant for those who died from other organ failure.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Factors associated with the use of supportive measures for palliative home care in the palliative subset population (n = 32,075).

https://doi.org/10.1371/journal.pone.0213731.t003

People who were married (OR = 1.75[1.65–1.86]), living together (OR = 1.34[1.18–1.53]), forming a one-parent family (OR = 1.27[1.14–1.41]), or being part of another sort of household (OR = 1.35[1.15–1.58]) were more likely to use at least one of all supportive measures than those living alone. People living in the Flemish (OR = 1.33[1.20–1.46]) or in the Walloon region (OR = 1.15[1.03–1.27]) were more likely to receive at least one of all measures than those living in the Brussels Capital region. People living in an area with a lower degree of urbanisation were more likely to use a supportive measure than those in an area with a higher degree of urbanisation.

Educational level was found to be a significant predictor of supportive measure use only in the group with post-secondary school education, and only for the use of a non-statutory measure (OR = 1.22[1.10–1.36]) or for any supportive measures (OR = 1.13[1.00–1.27]). Having higher standard housing is a significant predictor of statutory measure use, of any measure use, but not of non-statutory measure use. Nationality was included in the analyses but was found not to be significant.

Strengths and limitations

To our knowledge, this is the first study to measure uptake of both statutory and non-statutory supportive measures for palliative home care on a population-level. In Belgium, membership to one of the seven national health insurers is mandatory for the full population, thus the data cover (nearly) all individuals. However, additional private insurance is possible, which is not included in our data. We were able to measure multiple types of supportive measures, such as financial incentives, supportive home care services and multidisciplinary support teams. The use of linked data for all Belgian citizens for a given year from routinely-collected administrative, disease-specific and insurance databases offers a rich, powerful and reliable resource for studying important public health issues [18,19]. Our data also has its limitations. Although the identification of people in the palliative subset is based on a previously validated and frequently used estimation, the estimation is based on underlying cause of death, not on an actually observed or stated palliative care need[14]. Reimbursement data also do not capture other forms of support for palliative home care (e.g. care given by informal caregivers, services organised outside reimbursement schemes) or social support (e.g. religious based, kinship networks).

Interpretation of results

Only a third of home-dwelling people with potential palliative care needs made use of a statutory supportive measure for palliative home care. This low uptake is possibly related to the legal criteria surrounding these measures; in order to be granted legal palliative home patient status, life expectancy must be estimated as ‘more than 24 hours and less than three months’ [17]. Firstly, prognostication of survival in patients with advanced cancer has been well-researched, showing consistently that both physicians and patients are unable to correctly predict their life expectancy, most commonly overestimating it [20–23]. The disease trajectories of non-cancer serious illnesses are even more unpredictable, further complicating the process of prognostication [24–26]. Secondly, people confronted with non-cancer illnesses like organ failure or dementia often see their illness as ‘a way of life’[27] rather than a life-threatening illness, therefore possibly not perceiving themselves as having palliative home care needs [28–30]. The possibility of receiving a statutory supportive measure can thus come too late or even not at all. This hypothesis is supported by findings of a previous study that focussed on the physician’s reasons for not referring people with life-limiting illnesses to specialist palliative care services in Belgium in which physicians mentioned ‘not [having] enough time to initiate palliative care’ as one of the three major reasons for non-referral [13]. Indeed, recent survey data from Flanders shows that specialised palliative care is initiated with a median of only twenty days before death [31]. Our finding that younger people and those who died of neoplasms were most likely to use statutory supportive measures confirms previous findings [11], and could be related to the problems of recognising the terminal phase, something frequently reported to be more difficult in older population groups with chronic illnesses other than cancer [32,33].

The uptake of non-statutory supportive measures was higher than statutory supportive measures. The criteria for these non-statutory supportive measures are based on high care needs combined with high out-of-pocket costs for two consecutive years, and not on life expectancy, which could explain why they are used more frequently than statutory measures [34–36]. This could also explain our finding that the chance of using a non-statutory measure was highest among people who died of COPD, who are prone to high care needs and out-of-pocket costs [37,38]. It remains unclear, however, whether statutory and non-statutory measures contribute equally to the effective support of palliative home care.

The uptake of supportive measures for palliative home care seems to be influenced, apart from aspects related to prognosis, by factors such as social support and socio-economic position. The higher chances of using statutory and non-statutory measures in those living with others compared with single-person households suggests that the presence of social support in the household is an important factor in organising palliative home care and using supportive measures to do so. The influence of socioeconomic characteristics on the knowledge or use of palliative care has very rarely been studied, and with differing results [39–41]. Our findings partly suggest a social gradient exists in the use of supportive measures; the lower likelihood of using non-statutory measures in the lowest income category can indicate inequality. This result differs from the uptake of statutory measures, however, where people in the highest income category were significantly less likely to use these measures when compared with the lowest income category. The better chances of people with post-secondary education receiving a non-statutory measure may reflect differences in social capital that give them an advantage in knowing about and obtaining the measures. The advantage of those living in a home with a high standard of housing compared with those with a lower standard is an additional indication that there is some social inequality.

It seems likely that knowledge of existing supportive measures also plays a role. Because palliative care in Belgium has traditionally focussed mainly on cancer patients [42], their advantage in using statutory supportive measures could be due to both oncologists and patients being better informed. We also found that people living in Flanders were more likely to use a supportive measure than those in the Brussels Capital region. This may be explained by regional differences in general information provision and knowledge of palliative care or to differences in the social environment (e.g. lower social fragmentation in more rural areas) and national healthcare policy (concentration of large academic hospitals in metropolitan areas creating a pull factor towards end-of-life care in hospitals) [43].

Suggestions for further research and policy recommendations

Our study is a first step in understanding who makes use of supportive measures for palliative home care, and who is missing out on them. Future research should focus on how inequalities in the uptake of these measures influence different outcomes, such as place of death, quality of care, and costs at the end of life. Additionally, differences in the impact of using statutory or non-statutory measures on these outcomes should be compared. In this way, public health policy can be quantitatively evaluated and further improved. Qualitative research also needs to further examine the motivation and reasons behind the differences in uptake between certain population groups, and the differences in uptake between statutory and non-statutory supportive measures, as these are not registered in administrative data.

The large gap in uptake between statutory and non-statutory supportive measures among a subgroup of people in potential palliative care need suggests that there remains inequity in accessing statutory measures. Policy-makers should focus on including underreached groups, such as older people and those with non-cancer illnesses. The life expectancy criterion to qualify for the statutory supportive measures should be re-evaluated. This is possibly a major barrier that prevents many patients with palliative care needs to make use of them, in particular older patients and those with a non-cancer diagnosis.

Data availability statement

Due to ethical concerns with regard to sensitive and potentially identifying data, the supporting data cannot be made openly available, as stated by the Sectoral Committee of Social Security and Health—Department Health and the Data Protection Authority. Data availability and access regulations are available upon request with the separate database administrators. The InterMutualistic Agency can be contact via the website (https://ima-aim.be), Statistics Belgium via e-mail (statbel@economie.fgov.be), Belgian Cancer Registry via e-mail (info@kankerregister.org).