Global costs, health, and economic benefits of scaling up treatment and imaging modalities for survival of 11 cancers: a simulation-based analysis

Abstract

Background:

In addition to increased availability of treatment modalities, advanced imaging modalities are increasingly being recommended to improve global cancer care. However, estimates of the costs and benefits of investments to improve cancer survival are lacking, especially for low-income and middle-income countries (LMICs).

Methods:

Using a previously developed model of global cancer survival, we estimated stage-specific cancer survival and life-years gained (accounting for competing mortality) in 200 countries and territories for patients diagnosed with one of 11 cancers (Oesophagus, Stomach, Colon, Rectum, Anus, Liver, Pancreas, Lung, Breast, Cervix uteri, Prostate) representing 60% of all cancer diagnoses in 2020–30. We evaluated the costs and health and economic benefits of scaling up packages of treatment (chemotherapy, surgery, radiotherapy, targeted therapy), imaging modalities (ultrasound, x-ray, CT, MRI, PET, SPECT), and quality of care to the mean level of high-income countries, separately and in combination, compared to no scale-up. Costs and benefits are presented in $US 2018 and discounted at 3% annually.

Findings:

For the 11 cancers studied, we estimate that without scale-up there will be 76·0 million (95% UI 73·9–78.6) global cancer deaths for patients diagnosed in 2020–30, with 70% of deaths occurring in LMICs. Comprehensive scale-up of treatment, imaging, and quality of care could avert 12·5% (95% UI 9·0–16·3) of these deaths globally, ranging from 2·8% (95% UI 1·8–4·3) in high-income countries to 38·2% (95% UI 32·6–44·5) in low-income countries. Globally, we estimate comprehensive scale-up would cost an additional $232·9 billion (95% UI 85·9–422·0) in 2020–30 (6·9% increase in cancer treatment costs) but produce $2·9 trillion (95% UI 1·8–4·0) in lifetime benefits, yielding a return of $12·43 (95% UI 6·47–33·23) per dollar invested. Scaling up treatment and quality of care without imaging would yield a return of $6·15 (95% UI 2·66–16·71) per dollar invested and avert 7·0% (95% UI 3·9–10·3) of cancer deaths.

Interpretation:

Simultaneous investment in cancer treatment, imaging, and quality of care could yield substantial health and economic benefits, especially in LMICs. These results provide a compelling rationale for the value of investing in the global scale-up of cancer care.

Introduction

Cancer is a large and growing global health concern, with 18.1 million new cancer cases and 9.6 million cancer deaths estimated to have occurred globally in 2018,¹ and the burden projected to increase by more than 20% in 2030 to 22 million cases and 11.7 million deaths.² This burden falls disproportionately on low-income and middle-income countries (LMICs), which account for approximately 80% of disability-adjusted life years lost to cancer, but only 5% of global resources spent on cancer care.³ Five-year net cancer survival levels in LMICs are much lower than the levels attained in high-income countries, but even in high-income country settings there is substantial variation in net survival.⁴ Reasons for differences in survival include variations in the effectiveness of prevention, screening, and treatment of cancer, with evidence suggesting that scale-up of pharmaceutical treatment, surgery, and radiotherapy interventions could produce major health benefits,⁵ and yield substantial economic returns on investment.⁶

In addition to improving the availability of treatment modalities, advanced imaging technologies are increasingly being recommended for optimal cancer care (e.g. computerized tomography [CT], magnetic resonance imaging [MRI], single photon emission CT [SPECT], and positron emission tomography [PET]), and the rising incidence of cancer worldwide suggests that imaging-related costs will likely grow considerably.⁷ Economic evaluation of the value of imaging in cancer care is thus an important consideration for policymakers across different income and geographic settings. While limited studies on the cost-effectiveness of specific imaging modalities in high-income countries exist,^8,9 an important limitation of economic evaluations is that results estimated for one country may not apply elsewhere.¹⁰ Similarly, there are few estimates of the effect of imaging on cancer survival.^11,12 Given the rising burden of cancer, suboptimal survival in many countries, and rising costs of care, there is a need to demonstrate health and economic benefits of investing in the scale-up of diagnostics, treatment, and healthcare services.

In this analysis, we estimate the costs and lifetime health and economic benefits of scaling up imaging and treatment modality packages on cancer survival, both globally and by country income group. These estimates can help policy makers in different contexts prioritize investments in health systems to improve cancer treatment and survival.

Methods

Overview

We extended a previously developed microsimulation of global cancer survival for 11 cancers which comprise 60% of global diagnosed cancer cases (Oesophagus, Stomach, Colon, Rectum, Anus, Liver, Pancreas, Lung, Breast, Cervix uteri, Prostate), ¹¹ adding a module on costs and long-term survival outcomes. The model estimates stage-specific cancer survival in 200 countries/territories, and the availability and survival impact of specific imaging modalities (ultrasound, x-ray, CT, MRI, PET, SPECT), treatment modalities (chemotherapy, radiotherapy, surgery, targeted therapy), and quality of care (health system and facility-level factors that account for residual differences in survival not explained by cancer stage or treatment and imaging availability) (see Figure 1, and Appendix pg 2–4 for more details).¹¹ The joint distribution of age and stage of diagnosed cancer incidence is also accounted for by cancer and country in the model. We used Bayesian hierarchical models with four levels (income group, geographical area, geographical region, and country) to synthesise all available model input estimates and generate prior probability distributions, allowing us to regularise the reported data and estimate priors for countries for which no data were available. We used these priors as initial sampling distributions and calibrated the model to country-specific 5-year net survival estimates from CONCORD-3 (see Appendix pg 3–4 for details). Using the calibrated model we simulated all 11 cancers in each country, allowing us to make estimates for countries/cancers for which no data were available. Due to insufficient data on underlying cancer incidence and stage distribution, especially regarding the total (i.e. diagnosed and undiagnosed) cases in each country, we focus our analysis on the survival impact of treatment and imaging modalities conditional on diagnosis and stage, and do not consider the potential benefits of imaging on screening or early detection. However, aside from screening, increasing the use of imaging can help to improve the quality of cancer treatment. In addition to aiding in determination of cancer staging and primary treatment, imaging is used to assess treatment response, to evaluate possible recurrence or progression of disease, and to guide treatment decisions.

Figure 1: Conceptual cancer treatment cascade.

This cascade accounts for multiple factors that affect cancer survival from diagnosis to completion of therapy. SEER = Surveillance, Epidemiology, and End Results. IAEA = International Atomic Energy Agency. IMAGINE database = IAEA Medical imAGIng and Nuclear mEdicine global resources database.

Using the microsimulation model of global cancer survival, we estimated the costs and benefits of six different packages of scale-up in which we improved the availability of imaging and/or treatment modalities, and quality of care to the mean level of high-income countries: 1) Imaging only — Scale up all imaging modalities (ultrasound, x-ray, CT, MRI, PET, SPECT); 2) Treatment only — Scale up all treatment modalities (chemotherapy, radiotherapy, surgery, targeted therapy); 3) Quality only — Scale up quality of care; 4) Treatment + quality — Scale up all treatment modalities and quality of care; 5) Traditional + CT + quality — Scale up traditional treatment modalities (chemotherapy, surgery, radiotherapy) and imaging modalities (ultrasound, x-ray) as well as CT and quality of care; 6) Comprehensive — Scale up all imaging and treatment modalities and quality of care. We also modelled different stages of comprehensive scale-up from 0% to 100% to provide more realistic estimates of the impact of scale-up over time.

We compared the potential gains from scaling up all imaging modalities vs all treatment modalities, and also the gains foregone from not including imaging as part of comprehensive scale-up (i.e. Treatment + quality [no imaging] vs Comprehensive). Scale-up is defined as achieving parity with the mean estimated level of availability in high-income countries. For imaging, this implies coverage rates much lower than in countries such as the US, as availability threshold priors were set based on high-income countries with relatively low coverage so as not to overestimate the thresholds needed to ensure availability (see Appendix pg 3).¹¹

Cancer survival and competing mortality

We simulated the number of (diagnosed) incident cancer cases in each country in 2020–30, based on UN population projections¹³ and estimated cancer incidence rates from GLOBOCAN 2018¹⁴ (Appendix pg 5). We used the UN probabilistic population projections (PPP) to estimate the number of individuals in the population in each country from 2020 onwards. For simplicity, we assumed that cancer incidence rates remained constant throughout 2020–30. In each iteration of the model we sampled a population trajectory from 2020–30 and sampled an incident number of cases in each year.

We simulated the clinical course of each individual cancer patient diagnosed between 2020–30 over their lifetime until they died (from any cause), accounting for (net) cancer survival and competing mortality risks based on country-specific lifetable projections. Time of cancer death was sampled from 10-year stage-specific relative survival curves estimated among SEER cases diagnosed in 2000–2016,¹⁵ adjusted for the modelled 5-year survival estimates for each country and strategy¹¹ (Appendix pg 5–7). We assumed that if the patient survived 10 years they would not die from the cancer. (Mortality reductions for patients diagnosed in 2030 may therefore occur through 2040 in the model.) Although this approach is expected to capture the vast majority of cancer deaths, some cancer mortality may be missed, such as those due to late recurrence of breast cancer.¹⁶ Lifetable projections were based on estimates for 1950–2100 from the UN World Population Prospects.¹³ Lifetables were available by 5-year age group and 5-year intervals. We used linear interpolation to interpolate mortality rates between single ages and years and converted the (interpolated) annual mortality rates to annual probabilities. Lifetables were not available for 14 countries in the model, so we imputed mortality rates from similar countries (Appendix pg 7–8).

Cancer treatment costs

We undertook a literature review to estimate the relationship between direct cost of cancer treatment and log per capita Gross Domestic Product (GDP). We expect human resource costs — generally a high proportion of healthcare costs — to vary with per capita GDP. A previous multicentre, micro-costing analysis of the cost of delivering childhood cancer care in three countries found that personnel and pharmacy represented two of the top three cost items in all centres, and although human resource costs comprised a variable proportion of total costs, the ratio of total treatment cost to per capita GDP was similar across all centers.¹⁷ We obtained 108 cost estimates for 17 cancers in 30 countries and estimated a regression model to predict cancer treatment costs as a ratio of per capita GDP (Appendix pg 8–12). Due to the scarcity of available cancer treatment costs, we used cost estimates for cancers not included in the model as they were the only estimates available for some countries (e.g. Non-Hodgkin Lymphoma in Malawi). We thus obtained cost estimates for 17 cancers. While some of these cancers are not included in the model, these estimates were useful for informing the costs of cancer treatment in areas such as sub-Saharan Africa where there are very few other estimates available. Most costs were reported in $US, but for any costs reported in local currency the year-specific conversion rate to $US was used. We then calculated the ratio of reported costs to the country’s estimated per capita GDP for the same year. Based on US Medicare estimates and data from other countries, we assume that imaging comprises 10% of total cancer treatment costs (Appendix pg 13–14). Although Medicare estimates are based on a specific population group (generally 65+), the majority of cancers in the US occur in this age group, and the use of Medicare reimbursement costs is common practice for estimating direct medical costs in economic evaluations.¹⁸ We assume that these costs include both operating costs (e.g. health professional salaries, equipment maintenance and licensing fees, etc.) and capital costs (e.g. construction and installation) which have been amortized into the reported treatment costs per patient. We do not include the costs of diagnostic work-up in our analysis. However, because we model all patients conditional on diagnosis these costs would cancel out in the incremental comparison to our baseline scenario as we assume that the number of diagnosed cancers is unaffected by the modelled policy interventions.

In the model we assume that if a needed treatment or imaging modality is not available then the patient incurs only 50% of the treatment and imaging costs, as estimates of non-procedure costs (e.g. hospitalization) account for roughly half of cancer treatment costs (Appendix pg 13). This assumption also helps guard against underestimating the costs of increasing the availability of treatment and imaging modalities in the counterfactual scale-up scenarios.

GDP estimates used in the model were based on International Monetary Fund data ($US 2018) for 2015–23¹⁹ which we projected to 2030 using the average percentage growth between 2015 and 2023. We applied a ceiling/floor of +/−8% growth in GDP to bound projections within historical and anticipated trends (see Appendix pg 14). GDP projections were truncated at 2030 values for future years. For scenarios in which we scaled up quality of care, to account for the complexity of implementing various quality improvement programmes, we also include a 20% cost increase for health system strengthening for countries below the high-income mean for quality of care, based on estimates of the cost of scaling up other targeted health programmes from earlier studies.^17,20

Productivity Estimates

To estimate the economic benefits of improving cancer survival we used the value-of-life-year approach (also called the full income approach), which recognizes the intrinsic societal value of a life-year, even if the person is not in the workforce. This approach was based a on a value of 2·3 times GDP per capita per year in low and middle-income countries, and 1·4 times in high income countries, following the methodology in the Lancet Commission on Global Health in 2035, which estimated the willingness-to-pay in different settings for a change in mortality rates associated with a one year increase in life expectancy.²¹

As a sensitivity analysis we used the more conservative human capital approach, in which the economic value of a life-year is based on a person’s formal economic contribution, operationalized using 1-times per capita GDP in our model.^22,23 We only accrued productivity benefits between ages 18 and 64 years in the model using the human capital approach to reflect typical working ages. Economic benefits in both the full income and human capital approach only accrue to individuals who are still alive in the model, thus accounting for differences in life expectancy as we simulate country-specific cancer and all-cause mortality.

Model Outcomes

For each strategy, we calculated the number of cancer deaths averted, life-years gained, cancer treatment costs, productivity gains, and return on investment, compared to a baseline scenario or status quo of no scale-up. All outcomes were calculated for patients diagnosed with one of the 11 modelled cancers in 2020–30. Costs are presented in $US 2018. We discounted life-years, costs, and economic benefits at 3% annually. We ran 1000 simulations of the model for all scenarios, sampling from the 100 best-fitting parameter sets identified by calibration, previously described.¹¹ We report the mean and 95% uncertainty intervals (UI), calculated as the 2·5 and 97·5 percentiles of the simulation results. The simulation model was developed in Java (version 1.8.0).

Role of the funding source

The funders of the study had no role in the study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all of the data and accept responsibility to submit for publication.

Results

In our baseline scenario (i.e. with current availability of treatment, imaging, and quality of care), we estimate that among patients diagnosed with one of the 11 modelled cancers in 2020–30 there will be 76·0 million (95% UI 73·9–78·6) cancer deaths globally. We find that over 70% of these deaths will occur in low-income and middle-income countries: 4·3 million (95% UI 4·0–4·5) in low income, 16·1 million (95% UI 14·0–18·2) in lower-middle income, 34·3 million (95% UI 33·1–35·4) in upper-middle income, and 21·2 million (95% UI 21·0–21·5) in high-income countries.

Improving the availability of treatment in low-income and lower-middle-income countries is estimated to avert more cancer deaths than investments in imaging modalities alone, while the reverse is true for upper-middle income and high-income countries, where improving the availability of imaging is estimated to avert more cancer deaths than improving the availability of treatment (Table 1). Improving quality of care alone yields a larger mortality reduction in low-income countries than imaging only, but yields the smallest impact in other income groups. However, we find that improving quality of care is a critical component of comprehensive scale-up.

Table 1.

Health Benefits among Cancer Cases Diagnosed between 2020 and 2030 Under Various Scenarios of Scale-Up (11 modelled cancers)

	Cancer deaths averted (95% uncertainty interval)		Projected life-years gained across the lifetime, millions (95% uncertainty interval)
	Number	Proportion of total deaths	Undiscounted	Discounted (3%)
Global
Imaging only	2,463,500 (1,154,900–4,073,900)	3·2% (1·6–5·3)	54·92 (25·15–91·40)	33·17 (15·18–54·93)
Treatment only	4,095,600 (1,632,300–7,093,400)	5·4% (2·2–9·1)	103·28 (40·37–184·19)	58·36 (22·71–102·73)
Quality only	617,400 (58,600–2,042,900)	0·8% (0·1–2·6)	14·90 (1·29–50·81)	8·43 (0·73–28·34)
Treatment + quality	5,369,100 (2,894,300–8,032,800)	7·0% (3·9–10·3)	134·96 (72·84–208·11)	76·13 (40·94–116·06)
Traditional* + CT + quality	6,057,300 (3,730,400–9,251,200)	8·0% (5·0–11·8)	152·85 (95·82–233·36)	86·02 (53·61–132·12)
Comprehensive	9,549,500 (6,677,800–12,743,800)	12·5% (9·0–16·3)	232·30 (157·29–311·30)	133·71 (91·94–179·03)
Low income
Imaging only	79,100 (8,400–223,800)	1·8% (0·2–5·3)	1·86 (0·21–5·64)	1·08 (0·12–3·27)
Treatment only	458,300 (67,600–1,148,100)	10·7% (1·5–26·6)	11·93 (1·65–30·17)	6·58 (0·93–16·55)
Quality only	128,600 (1,800–391,900)	2·9% (0·0–8·9)	3·06 (0·03–9·72)	1·73 (0·02–5·41)
Treatment + quality	965,200 (645,200–1,374,100)	22·3% (15·3–31·5)	24·72 (16·38–36·29)	13·70 (9·08–19·80)
Traditional* + CT + quality	1,349,500 (1,146,000–1,622,700)	31·1% (26·3–36·4)	35·27 (29·54–42·52)	19·46 (16·35–23·29)
Comprehensive	1,656,300 (1,405,300–1,940,700)	38·2% (32·6–44·5)	42·41 (35·80–49·84)	23·62 (20·08–27·55)
Lower-middle income
Imaging only	702,100 (207,200–1,677,900)	4·5% (1·2–11·0)	16·79 (4·75–43·10)	9·83 (2·86–24·43)
Treatment only	2,534,200 (672,900–4,778,900)	15·5% (4·6–27·0)	65·35 (17·11–122·43)	36·56 (9·65–68·61)
Quality only	391,200 (6,600–1,583,100)	2·4% (0·0–9·4)	9·70 (0·16–41·11)	5·48 (0·09–22·95)
Treatment + quality	3,196,800 (1,015,400–5,755,300)	19·5% (7·3–32·0)	81·95 (25·25–148·26)	45·91 (14·36–83·23)
Traditional* + CT + quality	3,620,300 (1,169,000–5,761,600)	22·2% (8·4–32·1)	92·80 (30·08–149·70)	52·07 (17·00–83·58)
Comprehensive	4,811,900 (2,385,600–7,081,100)	29·6% (17·0–39·3)	121·09 (57·65–179·46)	68·72 (33·63–101·66)
Upper-middle income
Imaging only	1,378,400 (422,300–2,963,800)	4·0% (1·2–8·4)	30·02 (8·57–65·59)	18·40 (5·22–39·62)
Treatment only	860,600 (290,800–2,064,100)	2·5% (0·9–6·0)	20·59 (7·02–49·55)	11·99 (4·01–29·33)
Quality only	63,500 (500–492,800)	0·2% (0·0–1·4)	1·43 (0·01–11·22)	0·81 (0·01–6·40)
Treatment + quality	930,300 (368,900–2,077,100)	2·7% (1·1–6·2)	22·16 (8·70–49·73)	12·89 (4·99–29·47)
Traditional* + CT + quality	881,500 (328,900–1,768,000)	2·6% (1·0–5·1)	20·38 (7·55–41·70)	11·88 (4·29–24·22)
Comprehensive	2,484,400 (1,386,300–3,657,200)	7·2% (4·1–10·4)	56·06 (30·25–81·73)	33·65 (18·21–49·52)
High income
Imaging only	303,900 (110,700–624,500)	1·4% (0·5–2·9)	6·25 (2·37–13·19)	3·86 (1·46–8·12)
Treatment only	242,400 (88,000–451,200)	1·1% (0·4–2·1)	5·41 (1·95–9·85)	3·23 (1·15–5·86)
Quality only	34,000 (100–148,600)	0·2% (0·0–0·7)	0·72 (0·00–3·09)	0·41 (0·0–1·8)
Treatment + quality	276,900 (128,900–513,500)	1·3% (0·6–2·4)	6·14 (2·91–11·42)	3·64 (1·76–6·74)
Traditional* + CT + quality	205,900 (52,700–465,300)	1·0% (0·2–2·2)	4·41 (1·16–10·09)	2·62 (0·66–5·94)
Comprehensive	596,900 (372,300–918,800)	2·8% (1·8–4·3)	12·74 (7·74–19·42)	7·72 (4·74–11·78)

^*Traditional modalities include chemotherapy, radiotherapy, surgery, ultrasound, and x-ray

Globally, the results show that while improving the availability of imaging modalities by themselves could enable 3·2% of cancer deaths to be averted (95% UI 1·5–5·2), imaging is a critical component of a comprehensive package of scale-up, as 12·4% (95% UI 8·7–15·7) of cancer deaths could be averted when imaging is also scaled up with treatment and quality of care, vs only 7·0% (95% UI 3·9–10·3) when there is no investment to scale up imaging.

We find that globally, as well as within each country income group, investing in imaging modalities has the highest return per dollar invested of the four scenarios assessed, due to the relatively smaller costs involved (Table 2). However, the absolute survival benefits in investing in imaging alone are also relatively smaller. Investing in treatment and quality of care yields higher survival gains (Table 1), but we find that the global return on investment can be doubled (from $6 to $12 per dollar invested) by also including imaging investments as part of a comprehensive package of scale-up (Table 2).

Table 2.

Economic costs and benefits among cancer cases diagnosed between 2020 and 2030 (11 modelled cancers) – All results compared to no scale-up and discounted at 3% annually

	Incremental cancer treatment costs (2020–2030), $ billion (95% uncertainty interval)		Lifetime return on investment: Full Income, (95% uncertainty interval)
	Difference	Percent increase	Productivity gains, $ billion	Net benefit, $ billion	Return per $ invested
Global
Imaging only	6·84 (1·77–15·86)	0·2% (0·1–0·3)	1,226·21 (540·05–2,161·8)	1,219·37 (535·47–2,157·29)	179·19 (84·71–625·09)
Treatment only	50·72 (14·92–111·88)	1·5% (0·8–2·4)	1,183·24 (504·9–2,206·54)	1,132·51 (489·13–2,114·69)	23·33 (12·4–60·4)
Quality only	169·17 (61·73–302·53)	5·0% (4·7–5·7)	136·75 (13·31–469·15)	−32·42 (−237·28–285·24)	0·81 (0·08–3·61)
Treatment + quality	225·50 (83·87–408·34)	6·7% (5·7–7·8)	1,386·07 (726·42–2,342·19)	1,160·56 (484·04–2,053·7)	6·15 (2·66–16·71)
Traditional* + CT + quality	198·56 (78·47–357·58)	5·9% (5·3–7·0)	1,302·58 (649·69–2,619·3)	1,104·01 (458·58–2,367·68)	6·56 (2·81–17·83)
Comprehensive	232·88 (85·92–421·97)	6·9% (6·0–8·0)	2,894·41 (1,794·55–4,025·16)	2,661·54 (1,631·20–3,775·64)	12·43 (6·47–33·23)
Low income
Imaging only	0·25 (0·15–0·39)	3·2% (2·0–4·5)	3·72 (0·44–11·35)	3·47 (0·14–10·98)	14·78 (1·56–39·24)
Treatment only	3·70 (2·36–5·16)	47·3% (29·8–61·9)	22·84 (3·52–56·71)	19·14 (−0·52–51·46)	6·17 (0·88–14·31)
Quality only	1·58 (1·15–2·09)	20·0% (20·0–20·0)	6·21 (0·06–20·98)	4·64 (−1·43–19·22)	3·94 (0·03–12·75)
Treatment + quality	6·02 (4·27–8·07)	76·8% (55·8–94·3)	46·98 (29·67–69·52)	40·97 (24·43–63·50)	7·81 (4·62–12·82)
Traditional* + CT + quality	5·56 (3·99–7·40)	71·0% (51·6–89·2)	66·56 (53·97–81·1)	61·00 (48·72–75·16)	11·97 (8·79–17·63)
Comprehensive	6·32 (4·53–8·48)	80·6% (59·5–98·3)	80·19 (67·67–95·23)	73·87 (60·83–88·56)	12·69 (9·42–18·24)
Lower-middle income
Imaging only	1·24 (0·52–2·25)	1·8% (1·0–2·6)	121·05 (34·17–303·19)	119·81 (33·36–301·03)	97·49 (29·93–239·15)
Treatment only	16·40 (5·31–33·50)	24·0% (8·3–47·5)	439·07 (106·80–839·42)	422·67 (99·57–820·90)	26·78 (13·04–52·78)
Quality only	13·73 (7·23–22·14)	19·4% (19·2–19·5)	66·19 (0·91–267·84)	52·46 (−15·64–253·02)	4·82 (0·07–20·46)
Treatment + quality	33·29 (15·89–57·33)	48·0% (29·4–76·2)	550·92 (173·79–1,013·68)	517·63 (151·60–976·72)	16·55 (6·81–34·13)
Traditional* + CT + quality	29·87 (14·32–52·74)	43·0% (24·7–68·2)	626·57 (213·16–1,027·21)	596·70 (191·85–996·41)	20·98 (10·11–40·83)
Comprehensive	34·77 (16·76–59·24)	50·1% (31·2–78·8)	826·02 (406·46–1,231·37)	791·25 (383·92–1,199·69)	23·76 (13·09–45·63)
Upper-middle income
Imaging only	3·13 (0·31–9·17)	0·6% (0·2–1·1)	815·03 (182·04–1,814·75)	811·91 (180·97–1,807·94)	260·64 (114·33–1,831·46)
Treatment only	14·33 (1·33–50·92)	2·7% (0·6–7·1)	448·12 (132·10–1,321·69)	433·79 (127·40–1,277·41)	31·26 (12·79–202·42)
Quality only	9·13 (1·13–19·29)	1·7% (1·6–1·8)	26·46 (0·12–162·4)	17·32 (−15·20–152·02)	2·90 (0·02–30·85)
Treatment + quality	23·80 (2·60–66·71)	4·4% (2·3–8·9)	476·65 (147·95–1,326·33)	452·85 (131·02–1,273·85)	20·03 (7·10–119·35)
Traditional* + CT + quality	14·16 (2·01–30·78)	2·6% (2·0–3·9)	410·51 (124·92–993·47)	396·35 (113·00–973·20)	29·00 (8·84–188·10)
Comprehensive	27·01 (3·10–72·27)	5·0% (2·9–9·2)	1,377·97 (614·90–2,204·39)	1,350·96 (603·89–2,194·41)	51·02 (19·61–391·94)
High income
Imaging only	2·22 (0·25–6·64)	0·1% (0·0–0·2)	286·40 (78·23–761·73)	284·18 (77·9–757·19)	128·86 (67·81–480·88)
Treatment only	16·29 (2·63–41·08)	0·6% (0·2–1·0)	273·20 (91·72–537·19)	256·91 (81·96–502·44)	16·77 (8·34–61·05)
Quality only	144·73 (52·22–260·20)	5·2% (5·0–5·4)	37·89 (0·02–204·28)	−106·85 (−240·28–54·13)	0·26 (0·0–1·54)
Treatment + quality	162·40 (58·15–291·05)	5·8% (5·4–6·3)	311·51 (120·33–581·47)	149·11 (−85·18–429·55)	1·92 (0·65–6·11)
Traditional* + CT + quality	148·97 (54·24–265·06)	5·3% (5·1–5·7)	198·94 (61·12–508·28)	49·96 (−151·78–374·44)	1·34 (0·32–4·93)
Comprehensive	164·78 (58·45–296·55)	5·9% (5·4–6·4)	610·24 (338·86–1,080·65)	445·46 (135·09–898·39)	3·70 (1·61–10·57)

^*Traditional modalities include chemotherapy, radiotherapy, surgery, ultrasound, and x-ray

Globally, we estimate that comprehensive scale-up would increase cancer treatment costs by about 7%, but with large variation by country income group: 80% increase in low-income countries, 50% increase in lower-middle-income countries, and about a 5% increase in upper-middle-income and high-income countries (Table 2). However, we find that the returns on this investment in scale-up would be substantial, worth nearly $2·9 trillion (95% UI 1·8–4·0) of gains globally over the lifetime of cancer survivors diagnosed in 2020–30, yielding a net benefit of $2·7 trillion (95% UI 1·6–3·8). This translates to a return on investment of more than $12 per $1 invested globally, ranging from nearly $4 in high-income countries to over $50 in upper-middle-income countries. We see that if comprehensive scale-up were to occur immediately in 2020, the benefits would outweigh the costs within a few years (Figure 2). While this level of scale-up is likely not feasible in such a short timeframe, our analysis of progressive stages of comprehensive scale-up reveals positive net benefits for all country income groups even at very low levels of scale-up (Figure 3), highlighting the potential benefits and short time to return of investments in cancer care. We also find that in lower-income settings, scale-up of traditional modalities (and CT) and quality of care has a substantial impact on cancer mortality reduction compared to comprehensive scale-up (31% vs 38% in low income countries and 22% vs 30% in lower-middle income countries), and also provides substantial returns on investment.

Figure 2: Cumulative net benefits 2020–2050 of selected scenarios by Income Group (Discounted 3%).

Shaded areas represent 95% UI. Vertical line indicates 2020–2030 period of new diagnosed cancer cases included. Full income approach used to estimate benefits. Note: Net benefits are not incremental (i.e. compared to baseline) in this figure.

Figure 3: Survival and net benefits of progressive comprehensive scale-up by Income Group.

Panel A: Proportion of cancer deaths averted (among cases diagnosed in 2020–2030)

Panel B: Net lifetime benefits compared to baseline (Discounted 3%)

Shaded areas represent 95% UI. Full income approach used to estimate benefits.

Using the human capital approach to estimate benefits yields lower estimates of net benefit, since this approach does not assign any benefit for life-years gained among adults aged 65 years and older, when many cancers occur (Appendix pg 15). However, with an estimated return on investment of $2·46 (95% UI 1·29–6·52) per dollar invested, we find that even with this much more conservative assumption of benefits gained there is still a substantial return. Country-specific results using both economic approaches are available in a public data repository [please insert margin link to Dataverse repository: (link will be provided before publication)].

Discussion

In this model-based analysis, we find that investments to scale up treatment, imaging, and quality of care for cancer could yield substantial health and economic benefits, especially in low-income and middle-income countries which have lower levels of cancer survival compared with high-income countries.^4,11

Currently, the net benefits of treating cancer in low-income countries are very small, largely because cancer survival is often so poor. However, we find positive returns on investment for each country income group across all stages of comprehensive scale-up, highlighting the potential benefits of even modest investments in cancer care in low-income countries. Indeed, investments that improve cancer survival can yield substantial long-term benefits for countries in all income groups, with global net benefits worth nearly $3 trillion if comprehensive scale-up were achieved for patients diagnosed and treated in 2020–30.

We find that investments to increase the availability of imaging modalities provide substantial returns, and indeed are a critical component of a comprehensive strategy to improve global cancer survival. Because treatment modalities are generally currently more available, the incremental impact of scaling-up imaging modalities is comparatively larger as imaging availability often has further to improve. However, the incremental impact of scaling-up imaging alone is generally much smaller than treatment, especially where treatment modalities are often lacking. For example, in low income countries we estimate that the survival impact of scaling up treatment modalities alone is about 6-times greater than scaling up imaging alone, and scaling up treatment and quality together yields survival impacts about 12-times greater than imaging alone. In contrast, in high-income countries the estimated impacts of scaling up imaging and treatment are roughly equal. Although barriers to cancer imaging in high-income countries are not substantial, they are often larger than those to treatment. However, the impact of improving the availability of either are relatively small since treatment and imaging are already generally available in these countries.

To our knowledge, this is the first global analysis to estimate the long-term survival impact and the resulting economic benefits of scale-up of imaging modalities for cancer. While earlier studies have assessed the comparative effectiveness and/or cost-effectiveness of specific imaging modalities for cancer, they often evaluated more proximate outcomes, such as change in clinical management.^8,9,24–27 Our analysis of lifetime health and economic benefits thus provides a compelling rationale for the value of investments in the scale-up of imaging modalities for cancer care, and especially highlights the role of imaging as a valuable component in conjunction with investments in treatment and quality of care. Indeed, our estimates of the impact of imaging may be conservative since we focus on the impact of imaging on survival conditional on diagnosis and stage, which does not account for improved screening and staging which can reduce treatment costs and improve quality of life in addition to survival.

However, there are limitations to our approach. For example, we assumed that cancer incidence rates remained constant within each country over the analytic timeframe of 2020 to 2030 for incident cases. Country-specific data on the prevalence of risk factors for cancer incidence, how they are projected to change over time, and their impact on cancer incidence would be needed to refine this assumption. This assumption may overestimate future cancer cases in countries with declining incidence (e.g. due to decreasing tobacco consumption), and underestimate cases in countries with increasing incidence (e.g. due to increasing obesity or aging population). However, as cancer incidence tends to change slowly over time — a study from the Global Burden of Disease estimated that the age-standardized incidence of all cancers increased from 296·09 to 306·75 per 100,000 between 1990 and 2017,²⁸ a 0·04% increase — we do not anticipate major shifts would occur within the modelled 10-year incidence period. The incremental nature of our comparisons to a baseline scenario also helps make our results more robust to changes in absolute incidence that may occur. Similarly, although our projections of per capita GDP are based on trends observed before the COVID-19 pandemic, our treatment costs and economic benefits are both based on per capita GDP, so our incremental results of net benefit are more robust to absolute changes in GDP that would impact all modelled scenarios.

We also assumed immediate scale-up in our counterfactual scenarios, which is not a realistic timeframe, especially as quality improvements are generally related to improved workforce performance or restructuring of services, which may take years to yield benefits. However, our analysis of progressive stages of comprehensive scale-up reveals positive returns on investment at all levels of scale-up. Also, because we simulate cancer survival curves over a 10-year period from diagnosis, improvements in simulated cancer survival occur over the decade following scale-up, thus accounting for some lag time between scale-up and averted deaths. In addition, we find that scaling up only traditional modalities (and CT) together with quality of care can provide much of the total potential improvements in lower-income settings, and may be more feasible as an initial area of investment, followed by gradual scale up of more advanced imaging modalities, as well as targeted therapy, for which the costs of all indicated use may be currently unaffordable even under concessional pricing agreements.²⁹

We also had limited data from countries on cancer treatment costs, especially imaging-related capital costs and training. As a result we were not able to model the cost of scaling-up individual treatment and imaging modalities, and assumed that costs were not stage-specific. Cost estimates that were available were for variable follow-up periods, and so may not include surveillance after completing treatment or management of recurrence. Further research on global cancer treatment costs and investments (e.g. detailed ingredients-based costing studies) would help to refine our approach and provide more precise estimates of the costs of scale-up in various contexts, especially as the cost ratio of products vs workforce or procedures vs hospitalization may vary by setting. For example, our assumption that non-procedure costs account for 50% of total costs is based on only two studies, both of which are from high-income countries (studies from LMICs were not available) and report estimates between 40% and 49%. Cancer-related cost estimates of health system strengthening and quality improvement efforts are also lacking. We base our estimates of 20% on empirical evidence for health system strengthening costs related to scale-up of treatment and prevention services for HIV/AIDS in low-income countries,²⁰ but cancer-related scale-up may require smaller or larger relative investments, depending on the country. Similarly, although accounting for quality of care is important to control for health-system and facility-level factors not explicitly included in the model, this parameter is admittedly abstract, covering topics such as adequate laboratory and pathology diagnostics, infection control, nursing standards, coordination of care, etc. Empirical estimates of specific quality of care indicators would be useful to inform our model estimates, as this parameter is currently fit solely via model calibration due to lack of data.¹¹

Lastly, we only considered diagnosed cancer cases in this analysis, and thus did not take into account the potential impact of interventions (such as screening programmes and digital health information systems) on improved cancer detection. Along with the potential to detect cancer cases at earlier stages, such improvements may lead to a larger number of diagnosed incident cases in settings where many cancers currently go undiagnosed — an increase for which health systems should prepare.

Nevertheless, our model incorporates parameter uncertainty around the inputs and their impact on cancer survival,¹¹ and we find that our estimates are likely robust to changes in these assumptions. In our sensitivity analysis in which we used much lower estimates of economic benefit with the human capital approach, we still find a global return of $2.46 per $1 invested. Thus, even if the costs of scale-up were twice as high as we assume, there would still be a positive return on investing in imaging, treatment, and quality of cancer care purely in terms of increased productivity among cancer survivors, not to mention the broader societal value of life-years gained as approximated by a full income accounting. In addition, investments in improving the availability of treatment and imaging modalities and quality of care would yield benefits for other cancers that are not included in our model. Hence, projected benefits are likely to be larger than that estimated. Furthermore, we do not consider the impact of improved imaging and quality of care on cancer screening outcomes, which could further improve survival and reduce treatment costs by detecting cancers at an earlier stage. As a result, our estimates of benefit are likely conservative. We find that even after comprehensive scale-up of treatment, imaging, and quality of care there still exists a survival gap between high-income and lower-income countries due to higher stage at diagnosis,^11,12 highlighting the importance of prevention and screening as critical components of comprehensive cancer control efforts.

In our model-based analysis of global cancer survival, we find that investments in imaging can yield important gains for countries in all income groups, and that imaging is a necessary component of a comprehensive strategy to improve the continuum of cancer treatment. Therefore, in addition to expanding the availability of treatment and improving quality of care, successful scale-up of high-quality cancer care worldwide will require investments in different imaging modalities and specialist human resources (e.g. radiologists, radiology technicians, nuclear medicine physicians, physicists, scientists, engineers, informatics professionals) to ensure well-maintained and well-calibrated imaging equipment and protocols, as well as access to radiopharmaceuticals.³⁰

An important finding of our study is the substantial benefit from comprehensive scale-up, with simultaneous investments in treatment, imaging modalities, and quality of care. It is important, therefore, to ensure integrated planning and investment that involves different departments and functions in the health system (e.g. chemotherapy, surgery, radiotherapy, imaging, care delivery), with special attention paid to human resources to ensure that staff are appropriately trained. Although these investments would require new funding, with large increases needed in lower-income countries as a proportion of current cancer treatment costs, the resulting improvements in survival and productivity gains would yield even greater benefits than the costs required, suggesting that such investments would provide substantial returns on investment while averting millions of deaths from cancer.

Research in context.

Evidence before this study

There are few estimates of the cost-effectiveness and survival impact of specific imaging modalities for cancer care. Furthermore, existing estimates are typically available for high-income countries only. We searched PubMed for studies on the costs and benefits of global scale up of cancer imaging using the search terms “cancer”, “imaging”, “costs”, “benefits”, and “global” on Aug 3, 2020, without language or publication date restrictions. We found no global economic evaluations of cancer imaging modalities.

Added value of this study

Using a microsimulation model of global cancer survival this study provides estimates of lifetime cancer survival and economic costs and benefits for patients diagnosed with one of 11 cancers (Oesophagus, Stomach, Colon, Rectum, Anus, Liver, Pancreas, Lung, Breast, Cervix uteri, Prostate) in 2020–30. We assess the costs and lifetime health and economic benefits of scaling up packages of treatment and imaging modalities in different country income group contexts, accounting for quality of care.

Implications of all the available evidence

For the 11 cancers, we estimate that there will be more than 75 million global cancer deaths for patients diagnosed in 2020–30 without scale-up, with 70% of these deaths occurring in low-income and middle-income countries. Comprehensive scale-up of treatment, imaging and quality of care could avert more than 12% of these deaths globally, ranging from 3% in high-income countries to nearly 40% in low-income countries. Globally, comprehensive scale-up would cost an additional $230 billion in 2020–30 (a 7% increase in cancer treatment costs) and produce nearly $3 trillion in lifetime benefits, yielding a return of over $12 per dollar invested. In contrast, scaling up treatment and quality of care without investments in cancer imaging would yield a lower return of just $6 per dollar invested, and avert only 7% of cancer deaths. Imaging is thus a critical component of a comprehensive approach to improving cancer care, and investments in imaging scale-up would yield substantial health and economic benefits, especially in low-income and middle-income countries.