VoxEU Column COVID-19

Fatal underfunding? Mortality in Spanish nursing homes during the first wave of COVID-19

The COVID-19 pandemic has had a disproportionate impact on older Europeans living in nursing homes. This column finds evidence consistent with a 'fatal underfunding hypothesis', suggesting that the regional variation in nursing home fatalities during the first wave of the pandemic in Spain is associated with indicators of underfunding such as understaffing, larger nursing homes, and occupancy rates. Coordination failures both between healthcare and long-term care and between central and regional governments also contributed.

Older individuals are more vulnerable to COVID-19. In Spain, for example, 87% of COVID-19 deaths in Spain have been among individuals aged 70 years and above1 and a significant share of COVID-19 fatalities have occurred in nursing homes. Also in Spain, 13% of all nursing home residents died from COVID-19 in the first wave of the pandemic. This figure rises to 22% of nursing home residents over the age of 80 (Costa-Font et al. 2021).

Not all countries were equally prepared to face the healthcare consequences of the first wave of the COVID-19 pandemic. Before the pandemic, hospital admissions were slower among nursing homes residents than equivalent populations in the community (Lee et al. 2003, Robins et al. 2013). Furthermore, community infections in Spain do not correlate with nursing home infections (Costa-Font et al. 2021). What can explain the high rate of nursing home fatalities? 

Austerity and underfunding 

Spain, alongside other Southern European countries, was heavily exposed to austerity cuts in the 2008–2013 period, which entailed a significant reduction available funding for long-term care (Costa-Font et al. 2018) and might have magnified the effects of the pandemic in nursing homes.  

Underfunded nursing homes can lack the sufficient human resources to perform their duties to the expected level of quality. A common reaction to underfunding has been a reduction in the hiring of permanent staff.  Underfunding has also led to larger nursing homes, which made it challenging for high-occupancy facilities to find rooms available for social distancing. 

One way to test for the underfunding hypothesis with the limited data available is to examine the association between proxy measures of nursing home underfunding, such as understaffing (the ratio of staff to places) and nursing home fatalities (relative to excess deaths). Other proxies include average nursing home size (larger nursing homes might not guarantee access to protective equipment) and nursing home occupancy rates (as occupancy plays a role in limiting the availability of spare rooms for self-isolation). Figure 1 shows some evidence of the regional variation in COVID-19 fatalities as share of excess deaths and nursing home occupancy rates. 

In Costa-Font et al. (2021), we show evidence of an association between nursing home deaths and excess deaths, with larger nursing homes exhibiting higher fatalities relative to excess deaths. Similarly, we find a reduction in relative fatalities per additional staff per place in a nursing home. However, these estimates come from a small number of observations and report an adjusted association that cannot be interpreted as causal.

Figure 1 Average occupancy rate (number of users/number of beds) and COVID-19 nursing home deaths by region


Note: Data of users (IMSERSO): Aragón, Canarias and Extremadura, data from 2016. Galicia, data from 2017. Rest of regions 2018.
Source:  RTVE and IMSERSO

Coordination failures

Other explanations, as discussed in a previous Vox column (Costa-Font et al. 2017), refer to limited investment and coordination of health and long-term care, giving rise to the well-known ‘bed blocking’ problem, which is deemed the cause of a significant share of excessive hospital care use.  Healthcare and long-term care services in Spain have traditionally been subject to several types of ‘coordination failures’ both between healthcare and social care services and between different levels of government.  Coordination failures can explain why, on average, most regions took between 26 and 31 days to report a case. 

Coordination plans for health and social care have been limited to a few regions. Only eight out of 17 regions in Spain had developed health and social care coordination plans at the time of the pandemic. The Spanish government made coordination failures worse after centralising healthcare stewardship, and social care remains uncoordinated in the hands of regional authorities. Finally, in some regions, older patients were refused emergency healthcare by major hospitals, with clinical guidelines explicitly saying not to admit older patients residing in nursing homes.

Concluding remarks

Underinvestment in nursing home care, and specifically regions exhibiting lower ratios of staff to nursing home places, is correlated with higher nursing home fatalities relative to excess deaths. Coordination failures make things worse. When evaluating long-term care programmes, it is important to bear in mind the externalities that they engender in other areas of economic activity such as healthcare, especially during a pandemic. 


Costa-Font, J (2018), "Subsidising long-term care: Lessons from subsidy expansions and cuts", VoxEU,.org, 4 October.

Costa-Font, J, E C Norton and L Siciliani (2017), "The future of long-term care: a public policy assessment of funding and organisation",, 12 September.

Costa-Font, J, S Jimenez Martin and A Viola (2021), “Fatal Underfunding? Explaining COVID-19 Mortality in Spanish Nursing Homes”, Journal of Health and Aging (in press). 

Lee, D T F, J Woo and A McKenzie (2003), “A review of older people’s experiences with residential care placement”, Journal of Advanced Nursing 37(1): 9-27.

Robbins, I, A Gordon, J Dyas, P Logan and J Gladman (2013), “Explaining the barriers to and tensions in delivering effective healthcare in UK care homes: A qualitative study”, BMJ Open 3(7): e003178.


1 Source: Spanish National Network for Epidemiological Monitoring (, and

1,050 Reads