The impact of health system crises on glioblastoma management and outcomes

During the past 15 years, public health systems have had to face two huge challenges, the global financial and the COVID-19 pandemic crises. The way European governments have dealt with these crises has been diverse. Austerity measures were applied during the financial crisis, whereas maintaining investment or even boosting public spending was the governments’ response during the COVID-19 crisis. Recently, it has been well established that fiscal austerity increases the prevalence of multimorbidity and worsens the health-related quality of life,¹ although the impact on excess mortality is less clear.^2,3 In contrast, the COVID-19 crisis resulted in a cumulative excess of deaths worldwide⁴ and increased stress on health systems. However, the impact these tensions and divergent policies have had on a specific and demanding disease such as Glioblastoma is as yet unknown.

For this purpose, we analyzed our histologically proven and systematically collected database of glioblastoma patients (WHO 2007 and 2016 diagnostic criteria, excluding secondary glioblastomas) during the period 2008–2021 to determine the impact of these crises on glioblastoma management and patient outcomes.

To contextualize the analysis, our public University Hospital provides and is the reference center for highly specialized attention to 1.5 million citizens. We split the analysis into three-period groups: First, a No Crisis Group (2008 to 2010, and 2015 to 2019 years); Second, a Financial Crisis Group (2011–2014) determined by when the austerity measures were directly applied to our regional health budgets, meaning on average an 8.2% cut (with regard to the previous 3 years) to the global health expenditures of the region⁵; and finally the COVID-19 Crisis Group (2020–2021).

We analyzed the main demographic; clinical characteristics of patients as well as their treatment, focusing particular attention on a primary indicator (survival rates) and several secondary indicators (duration of symptoms before the first image demonstrating the brain tumor, the time between this first neuroimage to the histologic diagnostic, and time from this first image to the treatment onset, surgery for partial and gross total resections and radiotherapy in case of biopsies). To determine independent factors related to survival, a Cox Regression analysis (backward stepwise method) was performed. All the well-established prognostic factors (age, Karnofsky status, surgery extension, and complementary treatment) together with the times periods of interest and the secondary indicators were entered into the model. The excess of mortality during the pandemic period was assessed comparing the monthly glioblastoma deaths regarding the mean monthly mortality from all previous periods.

Table 1 summarizes the results. Period crises show a significant worsening of almost all secondary indicators and a potential selection of the fittest patients as may be inferred from the increase in Karnofsky status percentages over 70 compared with the period without crises. Glioblastoma excess mortality peaks during COVID-19 pandemic followed the same pattern as the general population, although with a shorter duration (March, July, August, and November for glioblastoma patients with regard to March–May, July–August, and October–November for general population during 2020),⁶ and COVID-19 accounts for the direct cause of deaths in 8.1% of patients. In addition, periods of financial and COVID-19 crises present the worst and the best overall survival outcomes, although the differences were not significant. However, the multivariate analysis only identifies the financial crisis period as a significant independent prognostic factor for worst survival, in addition to all well-known prognostic factors.

Table 1.

Demographic, clinical, and treatment characteristics of patients with Glioblastoma diagnosed during 2008 to 2021, together with primary (survival) and secondary (times to first image, diagnosis, and treatment) indicators of its management and outcome. Patients are grouped according to the period of time analyzed (no crisis, financial, and COVID-19 crises). At the end of the table are the variables identified as independent prognostic survival factors in the Cox regression.

	No health crisis (n = 386)	Financial crisis (2011–2014) (n = 180)	Covid-19 crisis (2020–2021) (n = 114)	P value
Age (mean ± SD)	61.3 ± 11	60.5 ± 11.8	61.6 ± 11.5	1
Olders (>70 y)	21.2%	18.9%	28.1%	0.17
Sex (female)	37.3%	33.3%	29.8%	0.3
Symptoms until 1st image a	2 (1–4)	3 (1–6)	4 (1–8)	0.006
≥ 14 days	3.3%	7.5%	19.5%
Surgery
Gross total resection	39.2%	40%	35.1%	0.67
Partial resection	32.2%	38.9%	30.7%
Biopsy	28.6%	21.1%	34.2%
Nº diagnosis × year	48.3 ± 8.1	45 ± 9.4	57 ± 5.7	0.31
KPS ≥70	70%	79.4%	77.9%	0.035
Time to diagnostic b	22 (13–34)	24 (14–35)	19 (13–28.5)	0.89
Treatment
Palliative	21%	15.5%	14%	0.079
RDT	6%	10.3%	0%
CMT/RDTc	60.8%	59.8%	77.2%
Clinical trial	12.2%	14.4%	8.8%
2nd line treatments	65.8%	66.4%	62.5%	0.85
Time to treatment	25 (16–35)	30 (19–42)	22 (15.75–31)	0.028
Time to radiotherapy	38 (32–46.5)	42 (35–54)	34 (28–41.5)	<0.001
Overall survival	10.9 (3.9–19.1)	9.4 (3.9–18.4)	14.4 (4.4–15.3)	0.23
Cox regression	Variables	HR	CI 95%
	Age	1.03	1.02–1.04	<0.001
	GTR	0.3	0.23–0.38	<0.001
	Partial resection	0.41	0.32–0.52	<0.001
	Biopsy	1
	KPS ≥ 70	0.48	0.36–0.65	<0.001
	CMT/RDT	0.18	0.11–0.31	<0.001
	RDT	0.3	0.17–0.51	<0.001
	No treatment	1
	Economic crisis	1.34	1.17–1.67	0.01
	COVID-19 crisis	1.04	0.77–1.41	0.78
	No health crisis	1

The range is the Interquartile Q25–75. Medians and ranges are expressed in days; SD: standard deviation; y: years; KPS: Karnofsky Performance Status; ^aduration of symptoms (related to the brain tumor) until the first image was conducted; ^btime between the first neuroimage was conducted to the histologic diagnostic; ^cConcomitant and adjuvant chemotherapy (CMT) to radiotherapy (RDT) in Stupp or Perry schedules; Overall survival is expressed in months and the differences assessed by the log-Rank test; HR: hazard ratio; CI 95%: confidence interval 95%; GTR: gross total resection. For comparisons between groups were used log-Rank test for time dependent variables, ANOVA and Bonferroni as post-hoc test for other continuous variables, and Kruskall–Wallis test for categorical variables.

Politicians and health providers have to be aware that cuts in health budgets (in our case, decreases of over 21% in pharmacy, 11.1% in equipment, and 1.7% in Hospital Services)⁵ can potentially cost even more living time in cancer patients, than the sudden excess burden to health systems. Crises can be overcome through physician’s efforts and adapting protocols but the lack of resources cannot be amended.

Funding

None.

Conflict of Interest

None declared.

Authorship Statement

Conceptualization: J.B.; data curation: J.B and M.A.; formal analysis and methodology: J.B and F.A.T.; data acquisition: N.V, A.P, G.P., N.V., and M.M; writing—original draft: M.A. and J.B.; review and editing: F.A.T, N.V, A.P, G.P., N.V., and M.M.