What is the cost impact of second opinions in oncology? A retrospective review

Abstract

Purpose:

Second opinions in oncology may optimize treatment plans, resulting in improved outcomes such as prognosis and morbidity. Changes in treatment plans are often de-escalations in treatment intensity. However, data on the cost impact of changes in cancer treatment resulting from second opinions are limited.

Patients and Methods:

We used a cohort of 120 patients who presented to a high-volume cancer center for a second opinion and also had a documented first opinion—30 patients each from four disease types: colorectal, head and neck, lung, and myeloma. For the 43 total patients who had a change in treatment from the second opinion compared to the first opinion, we estimated costs of treatment for the first and second opinions across the modalities of surgery, radiation, and drug treatment.

Results:

Among the 43 patients with a change between the first and second opinions, 7 patients had a higher cost from the second opinion, 31 had a lower cost, and 5 had the same cost. Averaged across all 120 cases—43 with changes in treatment and the rest without— mean cost savings were $15,015 per patient, ranging by disease-type from $2,517 (lung) to $43,437 (myeloma). Decreases in cost were due to less-intensive surgery or drug therapy regimens, or shifts from treatment to no treatment (observation).

Conclusions:

Further investigation is needed across cancer types and in other settings to confirm the finding of de-escalations and resulting cost savings from second opinions.

Introduction

Second opinions in oncology may impact health care costs by affecting the utilization of cancer diagnostics and treatments. Prior evidence suggests that second opinions more often de-escalate care intensity than increase it,^1–4 raising the possibility that second opinions may reduce health care costs. Saving costs, where appropriate, is an increasingly important aspect of health care, both for patients seeking to avoid financial toxicity^5–7, as well as for self-insured employers and payors⁸. A mechanism by which second opinions could save costs is by providing expertise from subspecialists, whose more narrow-focus may increase their ability to risk stratify and determine which patients need more versus less aggressive management.⁹

In oncology specifically, most studies of the value of second opinions do not examine cost, and have been limited to single cancer disease entities, or single modalities e.g. pathology, radiology, or surgical consultation.^{2, 10–20} Several studies that examined pathology-only second opinions suggest diagnostic changes may reduce costs from the resulting avoidance of surgical procedures.^{21, 22} One study looking at found cost savings but was limited to lung cancer and did not describe cost methodology. ²³ And one non-peer reviewed study reported cost savings but included non-oncology second opinions and did not describe cost methodology.²⁴

To our knowledge, there are no published data with robust methodology examining the cost impact of treatment changes in oncology second opinions across multiple cancer types and across the modalities of surgery, radiation, and drug therapy. To assess this issue, we developed a costing methodology and analyzed cases from a previously published analysis of the value of second opinions in oncology.²⁵ That analysis found that some treatment changes were escalations in care intensity, but a majority of treatment changes were de-escalations in care intensity.

Methods

Study design:

This retrospective study used data from a previously published examination of treatment changes resulting from second opinion consultations performed at a high-volume cancer center with subspecialty expertise. Treatment changes were assessed by independent subspecialized clinician reviewers.²⁵ In the current study, we examined those patients with treatment changes from the first opinion to the second opinion that were expected to impact patient outcomes of prognosis and/or morbidity; those treatment changes involved adding, removing, or changing one or more of the three modalities for cancer treatment: surgery, radiation, and drug therapy.

Participants:

The cohort included 120 patients, 30 each with newly diagnosed, untreated multiple myeloma, colorectal, lung, or head and neck cancers, who presented sequentially for second opinions in 2018 to any provider within a given disease area. The overall sample reflected a consecutive cohort within each cancer type; any patient with a documented first opinion presenting for a second opinion to any physician within the four disease management teams was included. The first and second opinions were compared to determine which second opinions recommended a change in treatment that was expected to impact the outcomes of prognosis and/or morbidity. For example, switching from one drug to another with the same expected prognosis/survival and morbidity/quality of life was not counted as a change in treatment, whereas changing or eliminating a surgical procedure to achieve the same prognosis/survival but with less morbidity, or recommending palliative care instead of intensive treatment to achieve the same prognosis/survival but with less morbidity, were both counted as a change in treatment expected to impact outcomes. For 42 of the 120 patients (23–57% of the time, depending on the disease type), the second opinion recommended a change in treatment expected to positively impact prognosis and/or morbidity. For 1 patient (0–3% of the time, depending on disease type), the second opinion recommended a change in treatment expected to negatively impact morbidity, but not prognosis. ²⁵ These 43 cases were the basis for this analysis examining the difference in cost between the first and second opinions. Among the remaining 77 patients, there was one patient who had a delay in treatment due to receiving a second opinion, resulting in bleeding from his colorectal cancer requiring transfusion; since he ultimately received the same treatment after transfusion, he was not included in the analysis (but is included in Appendix Table 1).

Analysis:

Cost calculations were based on projected estimates using Medicare reimbursement rates rather than actual incurred costs, in order to allow standardized comparison of costs between first opinions and second opinions, regardless of how or where patients ultimately received treatment. For each of the 43 cases, clinician experts in the given disease area and modality (surgery, radiation, and drug therapy) who were not involved in treating the patient reviewed each case and confirmed specifics related to the first and second opinion treatments. These specifics included the exact surgical procedure(s); the modality of radiation (i.e. Conventional Radiotherapy [CRT], Intensity Modulated Radiotherapy [IMRT], or Stereotactic Body Radiotherapy [SBRT]) and the radiation fractionation and dose; the exact drug regimen, doses, and length of drug treatment. Estimated costs of treatment for the first opinion and the second opinion were then summed and compared. We examined the relationship between changes in cost and the type of expected improvement in outcomes, as well as with guideline concordance of the first opinion.

For each of the three cancer treatment modalities of surgery, radiation, and drug therapy, we aimed to include all costs seen as essential to the delivery of the initial course of treatment including supplies, equipment, and physician time, and to exclude any possible costs that were seen as not essential. We excluded outpatient clinic visits (Evaluation & Management visits) and any subsequent possible treatments that might result from cancer recurrence, complications related to treatment, etc. We used Medicare reimbursement rates across all modalities. Where applicable, costs were adjusted to 2024 dollars to account for inflation. The specific methods of assigning costs for each of the three modalities are described below.

Surgical Cost Calculation Method:

We used Diagnostic Related Groups (DRGs), which account for fixed amounts of Medicare reimbursement that hospital facilities receive for a given episode of surgical care. We matched the surgical procedures in each case description to a DRG using the CMS ICD-10-CM/PCS MS-DRGv33 Definitions Manual²⁶, which provides the DRG associated with each type of procedure. Once the DRG was assigned, the national average Medicare reimbursement rate was used to calculate the 2024 facility costs associated with each DRG.²⁷ DRGs include the costs of facilities, supplies, anesthesiology and pathology services, nursing, and other costs associated with a surgical procedure. As DRGs do not account for the professional (surgeon) fees associated with the recommended surgeries, we included this cost using the Professional Fees Ratio (PFR) for each DRG. PFRs are defined as the ratio of total payments to facility-only payments per admission.²⁸ To determine the estimated total cost of the recommended surgeries, we multiplied the DRG reimbursement rate by the PFR for that DRG.

We did not include any costs outside of DRG reimbursement, such as surgical consultation visits, pre-operative consultations and diagnostic evaluations, post-discharge care or outpatient medications. Only one outpatient surgical procedure was recommended for a participant. To calculate that cost, we added Medicare’s average reimbursement rate for the corresponding CPT code for both the facility and the physician.

Radiation Therapy Cost Calculation Method:

We identified the Current Procedural Terminology (CPT) codes associated with each course or radiation and confirmed with expert medical coders specializing in radiation coding and billing. In addition to the code for the actual treatment delivered, CPT codes for the additional standard procedures associated with treatment were applied. This included port films, weekly treatment management, complex treatment devices, basic dose calculations, physics check, clinical treatment plans, stereotactic treatment management, complex treatment delivery, and complex iso plan. Details of the CPT codes that were applied to each modality (CRT, IMRT, and SBRT) according to standard clinical practice are detailed in Appendix Table 2. For example, CPT codes for image guidance and complex treatment devices (i.e. custom immobilization devices) were assumed only to apply to IMRT and SBRT.

For each CPT code, we determined the frequency of billing over the entire RT course. The frequency of billing varied by the clinical context and technology utilized. For instance, the code for treatment delivery is billed every fraction, weekly treatment management is billed every five fractions, and clinical treatment planning is billed once in the entire course. To estimate cost, the 2024 Medicare reimbursement rate for each CPT code was obtained, adjusted based on the billing modifier, and multiplied by the number of times each code was billed. The costs assigned to each CPT code were then summed together to estimate the cost of each recommended RT course. Similar to the methods for the other modalities, we did not include the costs of radiation oncology consultation visits or common procedures performed before, during, or after radiotherapy (e.g. pulmonary function tests prior to lung RT, or gastrostomy tube placement during head and neck RT).

Drug Therapy Cost Calculation Method:

We used previously published methodology that accounts for drug costs, inclusive of drug acquisition, administration fees, and guideline-concordant supportive care medications, using Medicare reimbursement rates^{29, 30}. This method sometimes provides a range of the total treatment price for a regimen, because treatment costs were calculated under two sets of assumptions: 1) assuming aggressive supportive care medications, 2) assuming less aggressive and lower-cost supportive medications. The average between the two costs was used when they were different. The drug doses corresponded to a 70 kilogram body weight, 1.7 m2 body surface area, and normal renal function. The 2018 regimen costswere adjusted to 2024 dollars. We did not include any costs such as medical oncology consultation visits, laboratory draws associated with treatment delivery, etc. Prednisone costs reflected average oncology dosing.³¹ For therapies given until progression, we used median progression-free survival (PFS) time as the duration of therapy, from clinical trial data. ^32–44 For two thyroid cases requiring radioactive iodine treatment, we used mean payer negotiated price at NCI (National Cancer Institute)–Designated Cancer Centers. ⁴⁵

Results

Study Sample:

Among the 43 patients in this analysis, the mean age was 61.6, and 30% were women. There were no significant differences across these or other demographics between this group of 43 patients and the remaining 77 patients in our prior study who had no recommended change in treatment. (Table 1).

Table 1.

Patient characteristics

	Change in management from 2nd opinion expected to positively improve outcomes (n=43)	No expected positive change (n=77)	Total (n=120)
Mean (median) age at visit (years)	61.6 (63)	59.1 (59)	60 (60.5)
Female	13 (30%)	37 (48%)	59.1 (59)
Race
Asian-Far East/Indian	7 (16%)	8 (10%)	15 (13%)
Black or African American	2 (5%)	4 (5%)	6 (5%)
No value entered / Pt refused to answer	2 (5%)	4 (5%)	6 (5%)
White	30 (70%)	59 (78%)	89 (74%)
Other	2 (5%)	2 (3%)	4 (3%)
Ethnicity: Hispanic or Latino	1 (2%)	2(3%)	3 (3%)
Payer type
Medicaid	1 (2%)	7 (9%)	8 (7%)
Medicare	15 (35%)	29 (38%)	44 (37%)
Private	27 (63%)	40 (52%)	67 (56%)
Self-pay	0 (0%)	1 (1%)	1 (1%)
Primary language other than English	4 (9%)	5 (6%)	9 (8%)
Married / life/domestic partner	25 (58%)	49 (64%)	74 (62%)

Changes in treatment modality:

The proportion of patients with changes in treatment between the first and second opinion by disease-type is seen in Table 2. These changes in treatment involved changes within or to the combination of the three modalities of cancer treatment- surgery, radiation, and drug therapy. Among the three modalities, surgery was more often eliminated or reduced compared to the other two modalities: 6 (14%) involved adding surgery or increasing its extent, whereas 21 (49%) involved eliminating or reducing surgery. Changes in radiation and drug therapy were similar between additions/ increases versus eliminations/reductions. Nine (21%) cases involved a shift from treatment with one or more of the modalities to no treatment/continued observation.

Table 2.

Change in treatment from 2^nd opinion with expected impact on outcomes; change in diagnosis leading to change in treatment

	Colorectal cancer (n=30)	Head and neck cancer (n=30)	Lung cancer (n=30)	Myeloma (n=30)	Total (n=120)
Change in treatment with expected negative impact on morbidity 1	1 (3%)	0 (0%)	0 (0%)	0 (0%)	1 (0–3%)
Change in treatment with expected positive impact on morbidity2 and/or prognosis 3	7 (23%)	17 (57%)	11 (37%)	7 (23%)	42 (23–57%)
Number with change in diagnosis leading to the change in treatment with expected positive outcome 4	1 (3%)	5 (17%)	3 (10%)	4 (13%)	13 (3–17%)

¹Negative impact: One patient was recommended a chemotherapy regimen expected to negatively impact morbidity (compared to the first opinion) but not prognosis

²Morbidity: All 42 patients with positive impact on outcomes had improvement in short and/or long-term morbidity

³Prognosis: 11 of the 42 patients had improvement in prognosis (3 head and neck; 7 lung, 1 myeloma)

⁴This row is a subset of the previous row

Frequency and detail of cost increases or decreases between first and second opinions:

Appendix Table 1 shows the first and second opinion treatments and costs for all 43 patients with a change in treatment recommendation expected to impact outcomes. In total, 7 (16%) patients had a higher cost resulting from the second opinion change in management, 31 (72%) patients had a lower cost, and the remaining 5 (12%) patients appeared to have the same cost. Notably, all 5 cases with an apparent same cost nonetheless had de-escalations in care: Four of the five had less extensive surgery (e.g. hemithyroidectomy instead of total thyroidectomy; surgical resection with primary closure instead of free flap), but appeared to have the same cost because we used the same DRG billing. One of the 5 involved recommended observation of smoldering myeloma instead of clinical trial; both observation and clinical trial management were described as having zero cost. For the one patient who had a negative expected impact from the second opinion, change in treatment was recommended to a chemotherapy regimen that had a lower cost than the first opinion chemotherapy regimen.

Increases in cost and association with outcomes and guidelines-concordance:

The 7 patients with a higher cost of the second opinion included 3 colorectal patients and 4 lung patients. Among these 7, 3 had an expected improvement in prognosis resulting from the higher-cost management. These 3 were part of 11 total patients with an expected improvement in prognosis resulting from the second opinion. The other 4 with a higher cost of the second opinion had an expected improvement in short and/or long-term morbidity. Three of the 4 lung patients with a higher cost of the second opinion had first opinions that were not guideline concordant, out of a total of 5 patients with non-guideline concordant recommendations. The patient with the highest cost increase from the second opinion ($71,278) was one of the patients with a non-guideline concordant first opinion, which had recommended only palliative chemotherapy; the second opinion recommended curative-intent drug treatment, with expected improvement in the patient’s prognosis.

Cost Savings from Second Opinions:

Table 3 shows tabulated results of the cost of the first and second opinions and the resulting cost savings across the four disease types. The 43 patients with a change in management expected to impact outcomes had a total cost savings (inclusive of the 7 patients with cost increases) of $1,801,842, with an average savings of $41,903 per patient, with a range depending on the disease type from an average savings per patient of $6,866 (lung) to $186,158 (myeloma). When considered across the entire original sample of 120 patients who received a second opinion, the mean savings from second opinions was $15,015 per patient, ranging by disease type from $2,517 (lung) to $43,437 (myeloma).

Table 3.

Cost impact of second opinion changes in treatment recommendation

	Colorectal cancer (n=81 of 30)	Head and neck cancer (n=17 of 30)	Lung cancer (n=11 of 30)	Myeloma (n=7 of 30)	Total (n=43 of 120)
1st opinion total cost	$412,898	$786,449	$440,981	$1,328,058	$2,968,385
2nd opinion total cost	$326,917	$449,227	$365,450	$24,949	$1,166,543
Aggregate cost savings (difference between 1st and 2nd opinion)	$85,981	$337,222	$75,531	$1,303,109	$1,801,842
Average cost savings across patients with a change in treatment	$10,748	$19,836	$6,866	$186,158	$41,9032
Average cost savings, across all patients receiving a 2nd opinion	$2,866	$11,240	$2,517	$43,437	$15,0153

¹Includes one patient with a negative expected impact of the change in treatment

²Calculated as $1,801,842 ÷ 43, not as the average across the row

³Calculated as $1,801,842 ÷ 120, not as the average across the row

For colorectal, head and neck, and lung patients, most of the cost savings (the majority of patients as well as dollars saved) resulted from de-escalations in treatment intensity due to removing surgery from the modalities being used or from changes in drug therapy to less expensive regimens. For myeloma, most of the cost savings (the majority of patients as well as dollars saved) resulted from de-escalations in treatment intensity due to removing drug therapy and instead recommending observation (no drug therapy), for cases of smoldering myeloma or monoclonal gammopathy of unknown significance (MGUS). Some of these myeloma cases involved a change in diagnosis due to the second opinion, from active myeloma to smoldering myeloma or MGUS, leading to the recommendation for observation instead of treatment.

Discussion:

This study sought to understand the cost implications of second opinions from oncology subspecialists at a high-volume tertiary cancer center. Among the 43 cases in our analysis, 7 (16%) involved second-opinion changes leading to an increase in cost, 31 (72%) led to a decrease in cost, and 5 (12%) were found to have the same cost. Averaging the changes in cost across all patients within a given disease type who had treatment changes, there was a decrease in cost cost savings ranging from $6,866 to $186,158 per patient depending on disease type, with a mean of $41,903 among the 43 patients with treatment changes. Averaging the changes in cost across all 120 patients who sought a second opinion in our original sample, there was cost savings ranging from $2,517 to $43,437 per patient depending on disease type, with a mean of $15,015.

Notably, these results are in line with the only other known data reporting per-patient savings from second opinions in oncology. One of these studies of patients with lung cancer showed savings of $19,062 per patient.²³ The other non-peer-reviewed study showed savings of $4,306 per cancer patient, but did not describe the frequency or types of cancer.²⁴ Our findings add to this body of work demonstrating the cost impact of second opinions, across various disease types.

Our methods likely underestimate the cost savings from second opinions. We assumed site-neutral payments, i.e. that payor costs are the same at the first and second opinions. . Using Medicare reimbursement rates, this likely led to a large undercounting of savings for the majority of the sample with commercial insurance. DRG bundling also results in counting certain de-escalations in surgical care as having the same cost (e.g. the 4 cases described in the results), resulting in undercounting savings the majority with commercial insurance, where surgical costs are unbundled. Since many of the changes in treatment were surgical de-escalations, this undercounting of our total findings of cost savings is likely further magnified. Finally, given that the cases were reviewed in 2018 and cancer treatment options have continued to increase in cost since then, a present-day cohort would likely find further cost savings.

Our findings are novel for several reasons. First, this is the first published study to rigorously examine and describe the cost implications of second opinions in oncology across different cancer disease types, and across the full spectrum of cancer treatment modalities. Second, this study examined cost implications only from changes in management from second opinions that had an impact on a patient’s outcomes of prognosis and/or morbidity, thus avoiding overcounting the benefit of second opinions from changes that do not impact outcomes.

There are some limitations to this study. First, this analysis included a small cohort of patients, thus limiting our ability to draw stronger conclusions about cost savings for these cancer types or for other cancer types not included in the analysis. We also excluded patients who had been previously treated, limiting our ability to draw conclusions about the full spectrum of patients who present for a second opinion. Second, the estimated costs reflect a predicted cost of treatment based on the treatment plans described; the analysis did not assess if the second opinion treatment plan was followed or the actual cost of care. Relatedly, we described the costs only of the total first line treatment package of care based on the modality of treatment (surgery, radiation, drug), and did not account for the risks of cancer recurrence or surgical complications and other potential side effects from the treatment that can result in additional downstream costs, nor the added cost of the 2^nd opinion consultation itself. Third, although we attempted to develop a method for similar accounting of the costs across all three modalities used to treat cancer—surgery, radiation, and drug therapy—no method can be perfect in this regard. For example, we did not include costs associated with diagnostic testing related to treatment such as laboratory draws associated with chemotherapy, or ancillary services such as G-tube placement associated with radiation, which may have differed between the first and second opinion recommendations. Finally, while we limited the analysis to patients with expected changes in their outcomes, the patients’ actual outcomes are not known: we could not examine actual outcomes of all patients getting a second opinions, since some were treated elsewhere.

This research raises several avenues of further exploration. The findings suggest that, compared to other clinicians, subspecialists more often de-escalate the intensity of treatment, with associated cost savings, rather than escalating treatment to achieve better outcomes; this deserves additional scrutiny. We also found that de-escalation of treatment intensity often relates to surgical de-escalation or removal of surgery entirely from the treatment plan, which deserves further exploration, especially as non-surgical treatments evolve. We also found that many patients, especially those with pre-cancerous conditions or abnormal results, are de-escalated by subspecialists all the way to observation. This radical change in treatment plan greatly impacts both patient outcomes and cost.

This study found considerable cost savings from subspecialist second opinions in oncology for patients with a variety of cancer disease types. More research is needed to understand which patients would benefit the most from subspecialty care while ensuring broad access to this care, who would be appropriate to receive treatment in less subspecialized settings, and who would be appropriate for virtual second opinions to potentially save costs related to travel time and missed work.

Context Summary.

Key objective:

What is the cost impact of meaningful changes in recommended treatment between first opinions and subspecialist second opinions in oncology, using a standardized, multi-modality (surgery, radiation, drugs) costing framework applied across different cancer types?

Knowledge generated:

Among 43 patients whose treatment changed after a second opinion, costs decreased in 31 (72%), increased in 7 (16%), and were unchanged in 5 (12%). Among all 120 patients who presented for a second opinion, mean savings were $15,015 per patient, ranging from $2,517 to $43,437 depending on the cancer type. Reductions stemmed from surgical or drug de-escalation or observation rather than treatment.

Relevance:

Findings can inform referral pathways for oncology second opinions, payer benefit designs, and value-based models by highlighting where subspecialist review commonly de-escalates care intensity and lowers spend, supporting better outcomes at lower cost for patients.