The electronic databases searches yielded 4,798 articles, with another two identified through hand search. A total of 324 articles were excluded due to duplication. The title and abstract of remaining articles were screened and 4,398 articles were excluded because they did not meet the inclusion criteria. The full-text of the remaining 78 articles were then reviewed for eligibility, of which 31 were found to be eligible for inclusion. The PRISMA flow diagram provides detail of the screening process (Figure 1).
Characteristics of included studies
Table 1 presents the characteristics of included studies. The 31 studies were conducted in four different regions, including Asia (China, n = 10; Iran, n = 3; Thailand, n = 3; Turkey, n = 3; Vietnam, n = 2; and Malaysia, n = 2); Africa (Kenya, n = 2; Ethiopia, n = 1; and Morocco, n = 1); Middle East (Jordan, n = 1); South America (Brazil, n = 1); and Europe (Serbia, n = 1), with a multinational study exploring financial toxicity across Malaysia, Thailand, Indonesia, Philippines, Vietnam, Laos, Cambodia and Myanmar (19). The total sample size was 120,883, which ranged from 30 to 45692 participants. Majority of the participants were females (n = 65,564). The mean age of the participants was 57.7±7.8 years and ranged from 42 to 72 years. The majority of the studies focused on specific cancer types, such as: lung (20-24); breast (25-27); colorectal (28, 29); liver (30); ovarian (31); prostate (32); and stomach (33).
Measuring financial toxicity and health-related quality of life
Most of the studies used unvalidated self-designed questionnaires to measure the financial toxicity related to cancer diagnosis, treatment and care. Answers to questions, such as: “How much did you pay for the medical expense last month?”’; and “How much did you spend on the disease-related expenses other than medical expenses?” were often used to measure the financial toxicity during cancer treatment and care (21). One study applied a pre-existing generic financial assessment instrument, namely the PFW scale, which consists of: five items on the psychosocial; two items on financial resources; and one item on coping strategies (34). One study utilised the Chinese version of the cancer-specific Comprehensive Needs Assessment Tool (CNAT) (35).
Three instruments were used in six studies to measure the health-related quality of life (HRQoL) of cancer patients in general and disease-specific aspects of life (19, 21, 22, 31, 32, 34). The most frequently used HRQoL instrument was the Functional Assessment of Cancer Therapy (FACT). In particular, the FACT is a two-part instrument that assesses general HRQoL related to cancer and cancer therapy (FACT-G) and tumour-specific measures, such as prostate (FACT-P).
Another instrument that was often used in the assessment of HRQoL in cancer patients was the European Quality of Life Five Dimension (EuroQol/EQ-5D), which measured well-being in five dimensions: usual activities; self-care; pain/discomfort; anxiety/depression; and mobility (19, 22). The EuroQol/EQ-5D combines self-assessment with a valuation of quality of life in which full health is scored at “one” and death is “zero”. Two studies used the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30), which consists of 30 core items with five functional scales (cognitive, emotional, physical, role and social), three symptom scales (fatigue, pain and vomiting/nausea) and a global health and quality-of-life scale (22, 31).
Prevalence of financial toxicity
Three studies provided prevalence estimates (36-38) enabling a meta-analysis of the prevalence of financial toxicity from cancer diagnosis, treatment and care. The pooled prevalence of financial toxicity was 56.96% [95% CI, 30.51, 106.32] (see Figure 2). The random-effects meta-analysis showed that the pooled prevalence of financial toxicity among cancer patients varied from 17.73% [95% CI, 15.76, 19.94] to 93.38% [95% CI, 87.21, 99.99] in any cancer type after separating the data on rural and urban in one study (38). Rural dwellers had a substantially higher prevalence of financial toxicity estimates (93.38% [95% CI, 87.21, 99.99]). However, the heterogeneity in the ratio of prevalence was extremely high (I2 = 100%).
Mean medical costs
Table 2 presents the results of the mean estimates of cancer care costs using random-effects meta-analysis and sub-group meta-analysis. Medical costs were categorised into seven cost items: consultation; diagnosis; treatment, including surgery, radiotherapy, chemotherapy, hormone therapy, combined modalities and palliative/supportive care; inpatient care; outpatient care; and follow-up care. In total, 11 studies presented data on mean medical costs (21, 23-25, 27, 32, 33, 37, 39-41). Overall mean medical costs were $2,740.18, which ranged from $1,953.62 to $3,526.74 per cancer patient. Components of the overall mean medical costs included: $2,366.00 [95% CI, 1920.76, 2811.24] in any cancer type; $1,902.95 [95% CI, $-655.85, $4,461.74] in lung cancer; $4,961.80 [95% CI, $4,892.80, $5,030.80] in stomach cancer; $91.60 [95% CI, $72.87, 110.33] in breast cancer; and $6,141.30 [95% CI, $5,717.88, $6,564.72] in prostate cancer, with GDP per capita ranging from $858 in Ethiopia to $10,262 in China.
Three studies reported data on mean diagnostic costs (27, 32, 39). Expressed as random-effect estimates, mean diagnosis costs were higher for any cancer type ($138.90 [95% CI, $126.59, $151.21]; p < 0.00001), as well as breast cancer in women ($16.02 [95% CI, $15.12, $16.92]; p < 0.00001) and prostate cancer in men ($205.80 [95% CI, $168.32, $243.28]; p < 0.00001). Consultation costs significantly favoured higher medical costs (p < 0.00001) (39). The ratio of consultation costs to GDP per capita ranged from 1.77 to 2.16 in Kenya.
Mean medical costs also varied depending on the type of treatment modality. Mean medical costs of surgery were measured in three studies from Kenya (39), Vietnam (27) and Iran (32) with GDP per capita ranging from $1,817 to $5,506. The pooled mean medical costs of surgery were $1,678.80 [95% CI, $62.39, $3,295.20]; p = 0.04; I2 = 100%), which varied greatly from breast cancer ($82.35 [95% CI, $76.86, $87.84]; p < 0.00001) to prostate cancer ($3,709.50 [95% CI, $3,396.01, $4,022.99]; p < 0.00001). On the other hand, data regarding overall mean medical costs of radiotherapy were available in two studies (27, 32). A non-significant increase in total mean costs of radiotherapy favouring low medical costs burden was observed ($4,131.50 [95% CI, $-3,923.69, $12,186.69]; p = 0.31; I2 = 100%), with higher heterogeneity. The ratio of radiotherapy costs to GDP per capita ranged from 0.59 in Vietnam to 154.78 in Iran.
The sub-group meta-analysis of the total mean medical costs of chemotherapy favouring high financial toxicity were observed ($6,555.98 [95% CI, $-97.19, $13,014.76]; p = 0.05; I2 = 100%), showing an increase mean costs of: $476.48 per breast cancer patients; $1,372.50 per any cancer type; $10,540.00 per lung cancer patient; to $14,181.30 per prostate cancer patient. Two studies presented data on mean medical costs of combined surgery, chemotherapy and radiotherapy (23, 39), with total mean costs of $9,888.14 [95% CI, $-4,480.83, $24,257.12] and a substantial heterogeneity (I2 = 100%). One study reported that combined surgery and radiotherapy for any cancer type resulted in even higher associated mean medical costs ($1,749.35 [95% CI, $1,257.90, $2,240.80]; p < 0.00001) (39).
Mean medical costs of palliative care were measured in four studies from Kenya (39), Vietnam (27), Brazil (41) and Turkey (23) with GDP per capita ranging from $1,817 to $9,042. The random-effects meta-analysis estimated the total mean medical costs attributed to palliative care as $3,741.28 [95% CI, $2,241.19, $5,241.37]. Also, two studies conducted in Ethiopia (37) and Turkey (25) reported data on mean medical costs of outpatient care, which was significantly associated with higher financial burden ($673.03 [95% CI, $488.40, $857.66]; p < 0.00001; I2 = 85%). One study from Vietnam (27) with GDP per capita of $2,715 reported the mean medical costs of follow-up care in breast cancer patients as $356.24 ranging between $311.36 to $401.12 per patient.
Mean non-medical costs
Non-medical costs had two main subcomponents: direct non-medical costs; and indirect non-medical costs. Two studies conducted in Iran and Turkey with GDP per capita ranging from $5,506 to $9,042 reported quantitative data on mean non-medical costs (24, 32). The overall pooled mean indirect non-medical costs were $2,402.47 [95% CI, $-2,356.15, $7,161.09], with $17.34 [95% CI, $11.87, $22.80] per lung cancer patient and $4,873.93 [95% CI, $3,604.88, $6,142.98] per prostate cancer patient. However, there was high heterogeneity (I2 = 98%).
The total mean direct non-medical costs as reported by one study from Turkey were $334.00 [95% CI, $333.74, $334.26] per lung cancer patient (24). Mean direct non-medical costs were observed to be significant (p < 0.00001). Components of the direct non-medical costs included disease-related transfer, accommodation, informal and transportation costs. It was observed that mean transportation costs ($162.00 [95% CI, $125.307, $198.693]; p < 0.00001) were responsible for 48% of the total mean direct non-medical costs incurred by lung cancer patients (24). Also, informal costs were associated with significantly higher mean direct non-medical costs among prostate cancer patients, with mean costs of $2,454.70 ranging between $2,171.84 and $2,737.56 (p < 0.0001) (32). The ratio of informal costs to GDP per capita ranged from 39.44 to 49.72 in Iran.
Determinants of financial toxicity
Figure 3 presents pooled estimates of the determinants of financial toxicity. The sub-group meta-analyses showed that cancer patients with a household size of more than four were associated with a significant increase in financial toxicity (1.17% [95% CI, 1.03, 1.32]; p = 0.02; I2 = 0%). There was no significant heterogeneity among the three included studies (34, 42, 43). The meta-analysis revealed that cancer patients who received more than six cycles of chemotherapy were almost two times more likely to experience high financial toxicity (1.94% [95% CI, 1.00, 3.75]; p = 0.05; I2 = 43%). In three of the included studies (36, 37, 43), it was observed that cancer patients who attended private health facilities during the course of their disease were statistically associated with high-level financial toxicity (2.87% [95% CI, 1.89, 4.35]; p < 0.00001; I2 = 26%). One study indicated that prolonged length of hospital stay was significantly related to cancer patients encountering higher financial toxicity (1.88% [95% CI, 1.68, 2.11]; p < 0.00001) (43).
Using data from six studies (19, 34, 36, 38, 42, 43), the pooled estimate for health insurance as a determinant of financial toxicity among cancer patients was not a significant factor (1.19% [95% CI, 1.00, 1.42]; p = 0.06; I2 = 33%). However, according to the leave-one-out sensitivity analysis, the random-effects meta-analysis showed that not having health insurance was a significant risk factor for exposure to financial toxicity (1.29% [95% CI, 1.03, 1.61]; p < 0.03; I2 = 42%) when removing one study from China (38) from the pooled analysis. The sub-group meta-analyses indicate no statistically significant association with cancer-related financial toxicity by gender (0.97% [95% CI, 0.65, 1.45]; p = 0.89; I2 = 70%), stage at diagnosis (1.16% [95% CI, 0.79, 1.70]; p = 0.46; I2 = 32%), level of education (0.73% [95% CI, 0.27, 2.03]; p = 0.55; I2 = 97%) or income level (1.74% [95% CI, 0.68, 4.47]; p = 0.25; I2 = 97%).
Health-related quality of life burden
Seven studies reported data on illness cost and HRQoL (19, 21, 22, 31, 32, 34, 35). Random-effects meta-analyses showed that cancer patients experiencing high financial toxicity was significantly linked to low HRQoL (12.63 [95% CI, 9.04, 16.21]; P < 0.00001; I2 = 89%), with higher heterogeneity across studies (Supplementary Figure 1). The HRQoL areas significantly affected included physical well-being (11.05 [95% CI, 6.17, 15.93]; I2 = 13%; p < 0.00001); social well-being (13.49 [95% CI, 4.77, 22.20]; I2 = 82%; p = 0.002), emotional well-being (16.69 [95% CI, 4.17, 29.21]; p = 0.009; I2 = 87%) and functional well-being (13.22 [95% CI, 7.63, 18.81]; I2 = 7%; p < 0.00001).
Coping strategies for reducing financial toxicity
Five studies described different coping strategies that cancer patients and their families adopted to meet the costs of their illness (37, 44-47). These coping strategies can be described at four levels, namely: individual level; relationship level; community level; and societal level (see Figure 4). Coping strategies at the individual level included: using personal savings; selling assets; skipping bill payments; borrowing or incurring bank debt; and delaying/forgoing treatment (33, 37, 45, 47). On the other hand, strategies at the relationship level include: receiving financial support from family and friends; and emotional support from partners, friends and family members (44). Community level strategies commonly reported included obtaining financial assistance from workplaces, neighbourhoods, churches and non-governmental/charity organisations (37, 45). Strategies at the social level included: creating supportive policies, including a waiver to help cancer patients offset their medical bills; and promoting a pleasant social support environment, such as food, accommodation and transport for treatment programme (44, 45).
Quality assessment
Supplementary Figure 2 presents the results of the quality assessment of the included studies. Sixteen studies achieved an overall low risk of bias. Thirteen studies were rated as a moderate risk of bias often because there were no: identification of potential confounders; evidence of strategies to deal with effects of confounding factors; and/or description of statistical adjustment in data. Overall, two studies were rated as high risk of bias because of: outcome measurement; and statistical analysis issues. Outcome measurement issues were due to the use of unvalidated instruments and lack of clear definition and documentation of outcomes.