Bundling medical and pharmacy benefits: Impact on medical costs

Abstract

BACKGROUND:

Employers are increasingly seeking strategies to manage rising health care costs, including bundling or carving out specific benefits like pharmacy coverage. Although some studies suggest that integrating medical and pharmacy benefits leads to reduced medical costs, the mechanism remains unclear.

OBJECTIVE:

To evaluate the impact of bundling medical and pharmacy benefits on medical costs and utilization among self-funded employer groups and explore the potential role of population health management programs (PHMPs) as a mechanism for driving savings.

METHODS:

This retrospective, propensity score–matched, longitudinal cohort analysis used administrative claims data from a large health plan in Southeastern Pennsylvania. Self-funded group plan participants were included in the analysis if they had integrated medical and pharmacy benefits in both 2022 and 2023 (integrated group) or if they lacked integrated benefits in either year (comparison group). We additionally required that these plan participants did not experience major changes to their population health management benefits. Doubly robust models, adjusting for concurrent risk scores, were used to compare changes in per-member per-month (PMPM) medical costs and per-1,000-members per-year utilization between the integrated and nonintegrated groups. Propensity score matching was employed to create balanced comparison groups.

RESULTS:

Members with integrated pharmacy benefits experienced a statistically significant reduction of $32.48 PMPM (7.5%) in total medical cost growth relative to the nonintegrated benefits group. This reduction was largely driven by lower inpatient ($13.27 PMPM, 13.5%) and emergency department ($1.35 PMPM, 4.7%) spending. For members with a PHMP benefit, integrated pharmacy was associated with greater estimated savings ($39.28 PMPM, 9.0%) compared with those without integrated pharmacy. Additionally, members with integrated benefits were estimated to have a greater probability (10.9%) of outreach by a PHMP nurse compared with those without integrated pharmacy.

CONCLUSIONS:

Bundling medical and pharmacy benefits was associated with reduced medical cost growth, particularly in inpatient and outpatient settings. Coordination activities like PHMPs, enabled by data integration from bundled benefits, may be a key mechanism for achieving cost savings. These results support the consideration of benefit integration strategies for employers seeking to optimize health care spending and improve care management.

Plain language summary

Gray literature and peer-reviewed studies have shown that members of self-funded health plans that integrate medical and pharmacy benefits experience slower growth in medical costs compared with those that carve out pharmacy. We find a $39 per-member per-month savings for members of plans with bundled pharmacy coverage largely driven by population health management programs. When estimated separately, we still find a savings of $16 per member per month without a population health management program.

Implications for managed care pharmacy

This study provides new evidence for decision-makers that bundling medical and pharmacy benefits may have medical cost offsets increasing the value proposition over direct pharmacy benefit manager contracting.

With health care costs in the United States continuing their upward trend, evidenced by a 7.5% rise in total health care expenditures in 2023¹—almost twice the previous year’s growth—self-funded employer groups are particularly focused on strategies to manage these increasing expenses. Self-funded health plans for employers have been a growing choice to contain costs over fully insured alternatives. From 1999 to 2024, the proportion of total American workers covered by self-funded plans increased from 44% to 63%.² As part of the self-funded strategy, employers have utilized contracting directly with pharmacy benefit managers with the promise of lower drug spending.³

Self-funded group plans, when entering an administrative services contract with a health plan, can customize the menu of benefits and services the contract will cover. This may include behavioral and mental health, primary medical, pharmacy, and stoploss insurance. In theory, a third-party vendor can save money by “carving out” part of the benefits typically offered by a health plan, leveraging their expertise from specializing in the management of a single type of service or benefit.

There is a small but growing body of both academic and gray literature, literature not peer reviewed and disseminated outside normal publishing channels,⁴ that examines the impact of integrating medical and pharmacy benefits vs carving them out. Over the last decade, many major health plans have examined the impact of integrating medical and pharmacy benefits for self-funded employers. These studies find positive associations with pharmacy integration, particularly with respect to medical cost savings, with estimates ranging from $10 to $21 per member per month (PMPM.)^5,6

In a study using commercially insured, self-funded individuals covered by Cambia Health in Oregon, Washington, Idaho, and Utah, authors found that, after adjusting for member demographic characteristics, health plan benefit characteristics, risk scores, and chronic conditions, the cohort of members with integrated pharmacy benefits had medical costs that were approximately $12 PMPM lower than the cohort without integrated pharmacy.⁹ Similarly, in a propensity score–matched study of commercially insured, self-funded University of Pittsburgh Medical Center health plan participants in Western Pennsylvania, those in the integrated pharmacy group experienced $8.73 PMPM lower medical cost growth than the nonintegrated group.¹⁶ Studies also report reductions in inpatient hospitalization rates by as much as 15%, as well as reductions in emergency department (ED) visits by as much as 9% for groups with integrated medical and pharmacy benefits.^7,8

Peer-reviewed and gray literature studies have also found that savings associated with integrated pharmacy benefits are often higher for members with chronic conditions. For example, Lucas et al report savings of as much as $64 PMPM after 3 years.¹⁰ In a similar 3-year study, Smith-McLallen found that employers with integrated pharmacy benefits had an expected average annual savings per 1,000 employees of $35,528 across 5 common chronic conditions compared with a similar nonintegrated population.¹¹ Estimates published in gray literature have typically been larger. Aetna, for example, found savings up to $117 PMPM for those with chronic conditions.¹² Similarly, Blue Cross Blue Shield of Rhode Island found savings between $112 and $750 PMPM, varying by underlying chronic condition.¹³ However, neither study was subject to peer review.

Although medical cost savings are a common finding in both academic and gray literature, neither has provided convincing evidence of a mechanism that translates integrated benefits into medical cost savings. Several studies have posited that integrated pharmacy allows for the use of additional and/or more timely data on member health and behaviors.^9,10,16 Subsequently, these data can be used to inform risk stratification and other predictive modeling exercises and allow for better targeting for PHMP program outreach, although no consistent evidence of this has been presented. Integrated pharmacy benefits were associated with higher levels of PHMP participation in 2 studies^10,16 and lower levels in another.⁹

This study uses member-level data from Independence Blue Cross, a large insurer in Southeastern Pennsylvania, to evaluate differences in medical cost and utilization for self-funded employers with and without integrated medical and pharmacy benefits. Additionally, we explore population health management programs as a possible mechanism through which integrated medical and pharmacy benefits that lead to cost savings are generated.

Study Data and Methods

This study uses medical claims to estimate savings associated with integrated medical and pharmacy benefits. Our primary outcome measures include total medical costs, the sum of what was paid by both the plan and the member, and medical costs by service category. We also examine counts reflecting utilization of medical services. Cost and utilization data are aggregated to the person-year level and represented as PMPM level averages (ie, the sum of costs divided by months of active coverage) and services per 1,000 members per year (PKPY) (ie, the sum of services per month times 12, times 1,000). Aggregated data reflect either the baseline (2022) or follow-up (2023) year.

Demographic information about each member, including age, gender, and location, as well as product and benefit details, were compiled from enrollment records in addition to diagnostic cost group, DxCG,¹⁴ risk scores and indicators for several common chronic conditions. Cotiviti DxCG risk scores are the results of proprietary risk adjustment models that use demographics and medical claims to predict short-term and long-term medical cost risk.

This analysis considered self-funded employers with stable pharmacy benefit integration and population health management product status across both years of the study, 2022 and 2023. That is, if an employer had integrated pharmacy benefits and a PHMP buy-up in 2022, they were only eligible for inclusion if they also had those products in 2023. These inclusion criteria resulted in a combined total of 194 eligible employers, 101 with integrated pharmacy benefits and 93 without integrated pharmacy benefits.

For each of the 194 eligible employers, we identified all individual members attributed to them with at least 1 month of active medical coverage in each of the study years. The typical member meeting these criteria had more than 11 months of coverage in both 2022 and 2023. In addition to being an active member in both years, we required that members not be associated with more than 1 employer during the entire study window to ensure consistent exposure to the benefits structure of interest. Members needed to be aged between 19 and 63 years in 2022 so that concurrent dual enrollment in Children’s Health Insurance Program or Medicare was unlikely. Lastly, in rare instances of missing data, the member was excluded from the study. Figure 1 illustrates how the study samples were constructed. Additional details appear in a study diagram in Supplementary Figure 1 ^{(42.1KB, pdf)} (available in online article).

FIGURE 1.

Study Population

CAD = coronary artery disease; CHF = congestive heart failure; CKD = chronic kidney disease; COPD = chronic obstructive pulmonary disease.

STATISTICAL ANALYSIS

We conduct a longitudinal cohort analysis using doubly robust methods to estimate the impact of bundling pharmacy and medical benefits on outcomes, adjusting for concurrent risk. First, members in the integrated benefits cohort (treatment group) were propensity score matched to eligible members in the nonintegrated benefits cohort using a 1-to-1 “greedy” matching algorithm without replacement. Matching was conducted using data from the baseline year (2022) only. Members were matched on age, gender, product type (Health Maintenance Organization/Preferred Provider Organization [HMO/PPO]), concurrent medical cost risk scores (Cotiviti DxCG Model¹⁴), selected chronic conditions, including hypertension, diabetes, and mental and behavioral health, as well as asthma and chronic obstructive pulmonary disease (COPD), and lastly, we include exact matches on indicators for exposure to a population health management program buy-up and whether the member resides within the core plan service and contracting area in Southeastern Pennsylvania, New Jersey, and Delaware.

We then estimate cost savings using ordinary least squares (OLS) regression in an approach analogous to difference-in-differences analysis. We compare differences in year-over-year changes in outcomes between the integrated and nonintegrated benefit groups from 2022 to 2023. In this analysis, the coefficient estimate on the key variable of interest (the interaction between membership in the integrated benefit cohort, and an indicator for the year 2023) approximates the average difference in medical cost growth, over all covered lives, attributable to integrated pharmacy benefits over the 2-year period. Because large outliers can heavily influence point estimates in OLS, we re-estimate marginal effects with the right tail of costs truncated at $250,000. As an additional robustness check, we estimate 2-part models with risk scores discretized to quintiles.

Results

STUDY POPULATION CHARACTERISTICS

Table 1 shows the baseline characteristics of the study population before propensity score matching (PSM), and Table 2 shows after PSM. The intervention and comparison groups were well balanced on most covariates, with standardized mean differences less than 0.1 both prior to and after PSM, with values of most outcomes and covariates changing little. The largest preperiod differences prior to matching were observed in the proportions of PPO product, residing in-area, and having a population health management program buy-up. We find that the proportion of prematch members residing within the Independence Blue Cross core service area is significantly greater for the integrated benefits cohort (84.5%) compared with the nonintegrated pharmacy benefits group (60.0%)—a difference of more than 33% (P < 0.001).

TABLE 1.

Summary of Baseline Characteristics: Unmatched Population

Benefit group (sample size)	Unmatched
	Nonintegrated (N = 362,890)	Integrated (N = 196,300)	SMD
Cost outcomes (PMPM)
Total medical cost	$421	$423	0.001
Inpatient medical cost	$93	$100	0.004
ED medical cost	$33	$29	0.030
PCP medical cost	$22	$20	0.023
Specialist medical cost	$65	$66	0.001
Utilization outcomes (PKPY)
Inpatient visits	42.0	46.8	0.015
ED visits	182.4	200.4	0.025
PCP visits	1,702.8	1,701.6	0.001
Specialist visits	3,710.4	3,740.4	0.002
Demographics and product features
Active months medical coverage	11.28	11.19	0.038
Age, years	40.35	40.49	0.011
Male sex, %	49.1	44.1	0.101
Female sex, %	50.9	55.9
PPO product indicator, %	84.4	72.7	0.288
Concurrent risk maxa	2.27	2.39	0.021
Prospective risk max	1.93	1.97	0.009
Concurrent risk mean	1.55	1.61	0.016
Prospective risk mean	1.35	1.39	0.016
In-area indicator, %b	55.98	84.48	0.656
n in area	203,175	165,827
n out of area	159,715	30,473
PHMP indicator, %	43.27	70.87	0.581
n with a PHMP	157,023	139,118
n without a PHMP	205,867	57,182
Chronic conditions, %
Asthma	7.64	8.88	0.045
Anxiety	15.91	16.58	0.018
Bipolar	1.44	1.50	0.005
CKD	4.09	4.21	0.006
CAD	2.34	2.26	0.005
Depression	11.28	12.19	0.028
Diabetes	5.98	6.17	0.008
CHF	0.71	0.74	0.004
COPD	1.33	1.31	0.002
Hypertension	14.21	14.67	0.013

PKPY values are aggregations to the per 1,000 members per year.

PMPM values are aggregations to the per member per month.

^a Concurrent and prospective risk scores generated using Cotiviti DxCG models. ¹⁴

^b Indicator for whether a member lives within the Independence Blue Cross core service area.

CAD = coronary artery disease; CHF = congestive heart failure; CKD = chronic kidney disease; COPD = chronic obstructive pulmonary disease; DxCG = diagnostic cost groups; ED = emergency department (associated with visits that do not result in an inpatient admit); PCP = primary care provider; PHMP = population health management program buy-up; PKPY = per 1,000 members per year; PMPM = per member per month; SMD = standardized mean difference.

TABLE 2.

Summary of Baseline Characteristics: Matched Samples

Benefit group (sample size)	Matched: full sample			Matched: chronic condition sample
	Nonintegrated (N = 187,821)	Integrated (N = 187,821)	SMD	Nonintegrated (N = 83,865)	Integrated (N = 83,865)	SMD
Cost outcomes (PMPM)
Total medical cost	$443	$431	0.007	$678	$661	0.008
Inpatient medical cost	$101	$98	0.003	$169	$164	0.003
ED medical cost	$30	$29	0.003	$44	$43	0.008
PCP medical cost	$21	$20	0.007	$30	$30	0.001
Specialist medical cost	$69	$67	0.008	$95	$95	0.001
Utilization outcomes (PKPY)
Inpatient visits	49.2	48.0	0.006	72.0	72.0	0.001
ED visits	199.2	205.2	0.009	285.6	286.8	<0.001
PCP visits	1,687.2	1,724.4	0.011	2,463.6	2,500.8	0.014
Specialist visits	4,095.6	3,853.2	0.032	5,498.4	5,330.4	0.019
Demographics and product features
Active months medical coverage	11.24	11.18	0.025	11.45	11.40	0.024
Age, years	40.45	40.21	0.018	43.60	43.64	0.003
Male sex, %	45.24	44.81	0.009	40.31	40.05	0.005
Female sex, %	54.76	55.19		59.69	59.95
PPO product indicator, %	76.67	75.90	0.018	75.39	74.56	0.019
Concurrent risk maxc	2.45	2.43	0.004	3.64	3.66	0.002
Prospective risk max	2.03	1.99	0.009	2.86	2.82	0.007
Concurrent risk mean	1.68	1.64	0.011	2.54	2.51	0.006
Prospective risk mean	1.43	1.41	0.009	2.00	1.98	0.007
In-area indicator, %d	83.78	83.78	<0.001	83.92	83.92	<0.001
n in area	157,348	157,348		70,380	70,380
n out of area	30,473	30,473		13,485	13,485
PHMP indicator, %	69.56	69.56	<0.001	71.62	71.62	<0.001
n with a PHMP	130,648	130,648		60,064	60,064
n without a PHMP	57,173	57,173		23,801	23,801
Chronic conditions, %
Asthma	9.29	9.22	0.002	20.00	19.98	0.001
Anxiety	18.00	17.12	0.023	36.01	34.40	0.034
Bipolar	1.63	1.55	0.006	3.46	3.32	0.008
CKD	4.30	4.27	0.002	9.08	9.19	0.004
CAD	2.41	2.28	0.008	5.11	4.93	0.008
Depression	12.88	12.63	0.008	26.19	25.77	0.010
Diabetes	6.28	6.33	0.002	13.60	13.87	0.008
CHF	0.74	0.72	0.002	1.63	1.59	0.004
COPD	1.40	1.32	0.007	2.97	2.81	0.010
Hypertension	15.18	15.12	0.002	31.40	31.66	0.006

PKPY values are aggregations to the per 1,000 members per year.

PMPM values are aggregations to the per member per month.

^a Concurrent and prospective risk scores generated using Cotiviti DxCG models.¹⁴

^b Indicator for whether a member lives within the Independence Blue Cross core service area.

CAD = coronary artery disease; CHF = congestive heart failure; CKD = chronic kidney disease; COPD = chronic obstructive pulmonary disease; DxCG = diagnostic cost groups; ED = emergency department (associated with visits that do not result in an inpatient admit); PCP = primary care provider; PHMP = population health management program buy-up; PKPY = per 1,000 members per year; PMPM = per member per month; SMD = standardized mean difference.

To address the large imbalance in the percentage of members in the core service area counties between the unmatched treatment and comparison groups, we specified an exact match on an in-area indicator. Because the insurer has direct contracting with providers within their core service areas, costs for services provided by similar providers outside the core service area will likely be more expensive. The core service area is the geographic region in which the plan has direct network control including contracting, rate negotiations, value-based arrangements, etc. Thus, the plan likely has greater control over costs and cost growth in these areas when compared with outside. The second largest imbalance in the unmatched dataset is the proportion of members of group plans with a population health management program benefit. We address this imbalance by including an exact match on an indicator for having a PHMP benefit as well. After matching, the proportions of members with PPO coverage, with a PHMP benefit and residing in the core service area were comparable (Table 2). Additionally, although we did not explicitly match on medical costs, preperiod total medical costs did not differ between the intervention group and the comparison group prior to or after matching. We observe similar findings with measures of utilization and costs by service category.

MEDICAL COSTS AND UTILIZATION

Table 3 contains the results of the primary analyses. When compared with the matched nonintegrated cohort, members with integrated pharmacy benefits experienced $32.48 (7.5%) lower PMPM medical cost growth from 2022 to 2023 compared with matched members without integrated pharmacy benefits. These savings were largely driven by lower PMPM spending on inpatient admits, $13.27 (13.5%), and ED services, $1.35 (4.7%). Consistent with the existing literature, we examine the extent to which outliers are driving point estimates by top-coding all observed costs exceeding $250,000, a point where stop-loss insurance is often activated.⁹ We find that after limiting the distribution of medical costs to $250,000, the effect size is reduced but still substantial and significant—a savings of $22.10 (5.3%).

TABLE 3.

Regression Results: Full Matched Sample

Outcome	Estimatea	P value
Total medical cost PMPM	−32.48	<.001
Top coded at $250k (PMPY)	−22.10	<.001
With PHMP buy-up	−39.28	<.001
With PHMP buy-up and top coded at $250k (PMPY)	−24.54	<.001
Without PHMP buy-up	−16.84	.184
Without PHMP buy-up and top coded at $250k (PMPY)	−16.51	.067
Inpatient medical cost PMPM	−13.27	.013
Emergency department medical cost PMPM	−1.35	.038
Inpatient admits PKPY	−6.72	.632
Emergency department visits PKPY	−7.62	.020

PKPY values are aggregations to the per 1,000 members per year.

PMPM values are aggregations to the per member per month.

^a Regression coefficient for the interaction of post-period and integrated Rx benefit indicators using ordinary least squares controlling for period-average concurrent risk.

PHMP = population health management program buy-up; PKPY = per 1,000 members per year; PMPM = per member per month.

We also examine changes in utilization over the same period. Inpatient stays had a nonsignificant but directionally consistent reduction in the integrated cohort compared with those in the nonintegrated group by 6.72 (P = 0.632) fewer admits PKPY. Similarly, analyses of changes in ED visits showed that the integrated cohort had a statistically significant 7.62 PKPY (3.7%) decrease in ED visits compared with the matched comparison group.

These findings are consistent with the existing literature on integrated benefits. Although several papers have alluded to a mechanism, there has not been conclusive evidence as to why we observe these results. To explore this more fully, as well as examine the robustness of the main analytic specification, we explore potential drivers of savings for members with integrated benefits.

SENSITIVITY AND ROBUSTNESS ANALYSIS

Chronic Conditions

Several previous analyses have found greater savings in subpopulations with chronic conditions.^9,10,16 Although PSM has been used in assessing savings on chronic samples, to our knowledge, this is the first analysis to repeat the match for those with chronic conditions instead of filtering the matched cohort to those with at least 1 chronic condition. By performing matching on the chronic condition subsample separately, we are able to ensure that balance across all other covariates is preserved when we reanalyze members with those chronic conditions.

Table 1 shows the descriptive statistics of the matched chronic condition sample. Matching resulted in a sample of 83,865 pairs. We achieve balance across all covariates. In comparison with the unmatched and matched full data, the matched chronic sample is more likely to be older, female, and of higher concurrent and prospective risk, consume more services, and have 50% higher medical costs when compared with the full self-funded member population.

Among members with 1 or more of the selection chronic conditions, we find that the integrated pharmacy cohort experienced lower growth in total medical costs from 2022 to 2023 compared with the matched comparison group. Table 4 shows the results from these regressions. We find a $38.58 PMPM (P = 0.009, 5.8%) savings in total medical cost for the integrated pharmacy group driven by lower PMPM inpatient spending, $20.58 (P = 0.060), and a nonsignificant reduction in PMPM ED spending of $9.97 (P = 0.132). Like the primary analysis, we test the sensitivity of this finding to high-cost outliers. When the distribution of costs is top coded at $250,000, we also find a significant cost reduction of $25.33 (P = 0.011).

TABLE 4.

Regression Results: Chronic Condition Matched Sample

Outcome	Estimatea	P value
Total medical cost PMPM	−38.58	.009
Top coded at $250k (PMPY)	−25.33	.011
With PHMP buy-up	−54.50	.002
With PHMP buy-up and top coded at $250k (PMPY)	−22.39	.026
Without PHMP buy-up	1.78	.947
Without PHMP buy-up and top coded at $250k (PMPY)	−7.21	.644
Inpatient medical cost PMPM	−20.58	.060
Emergency department medical cost PMPM	−1.17	.376
Inpatient admits PKPY	−4.79	.081
Emergency department visits PKPY	−3.29	.594

PKPY values are aggregations to the per 1,000 members per year.

PMPM values are aggregations to the per member per month.

^a Regression coefficient for the interaction of post-period and integrated Rx benefit indicators using ordinary least squares controlling for period-average concurrent risk.

PHMP = population health management program buy-up; PKPY = per 1,000 members per year; PMPM = per member per month.

Population Health Management

As previously discussed, population health management program buy-ups have been posited as a potential mechanism through which pharmacy integration leads to medical cost savings. We explore this hypothesis by examining the effect of integrated pharmacy benefits for employers with and without a PHMP benefit buy-up. Because we specified an exact match on a PHMP benefit indicator, we can create matched subsamples by filter on the PHMP benefit indicator variable.

For those with a PHMP benefit, members with integrated pharmacy benefits had an estimated savings of $39.28 PMPM (P = 0.009) compared with matched members without integrated benefits. These savings equate to approximately 7.8% of total medical costs. Additionally, we re-estimated these figures with the same top-coding restriction on total medical cost at $250,000 as in the primary analysis. We find statistically significant savings of $22.39 PMPM (P = 0.002) which is approximately a 4.4% savings of total medical costs for the integrated group in 2022.

We conduct the same exercise for those that do not have a PHMP benefit. For those with integrated pharmacy benefits, we estimate a nonsignificant savings of $16.84 (P = 0.184) when compared with those without integrated pharmacy. When we re-estimate savings, with total medical costs top coded at $250,000 PMPY, we find a savings of $16.51 PMPM (P = 0.067) for members with integrated pharmacy benefits compared with those without. However, it is worth noting that, as the point of truncation is lowered from $250,000 to $150,000, the attachment point for many stoploss policies, we find a statistically significant savings and a nearly identical point estimate of $16.11 PMPM (P = 0.041).

To test whether the overall estimated savings associated with integrated pharmacy and medical benefits is driven by a PHMP benefit, we examine records for those identified for, and contacted by, PHM program nurse outreaches. For the matched cohort with a PMHP benefit, those with integrated pharmacy saw 12.8% more nurse outreaches than those without integrated pharmacy benefits (3.27% vs 2.90%, P < 0.001). Additionally, we employed logistic regression to estimate the likelihood of PHMP identification and nurse outreach conditional on age, gender, risk scores, product, chronic conditions, and whether the member had integrated pharmacy benefits. We find that, all else equal, those with integrated pharmacy were 0.314 (P < 0.001) percentage points more likely to be outreached by PHMP nurses than those without integrated pharmacy benefits. This reflects a 10.8% greater probability of outreach for those with integrated benefits compared with those without.

Two-Part Model

As an additional robustness check, we re-estimate the marginal effect of pharmacy benefit integration using a 2-part model.¹⁵ In these models, we transform concurrent risk scores by discretizing the variable into risk score quartiles to account for the long right tail of risk scores. We estimate the savings associated with integrated pharmacy benefits were estimated to be $47.28 PMPM (P < 0.001, 11.0%). In all, we find similar but larger effects when using the preferred 2-part model estimation; that is, our primary OLS estimates are conservative.

Clustered SEs

The decision to integrate or carve out pharmacy benefits is made at the employer level. Modeling savings at the member level does not account for potential employer-level variation that may drive such a decision. Although our specification controls for time-invariant employer effects, time-variant effects are not accounted for. We do, however, cluster SEs at the employer level and find that our primary estimate is still statistically significant.

Discussion

In our analysis, we examined the relationship between total medical costs and cost growth associated with integrated and without integrated pharmacy benefits. We find that, when compared with nonintegrated benefits, those with integrated pharmacy benefits saw lower total medical cost growth between 2022 and 2023, largely driven by lower inpatient and ED spend. These findings are consistent with the existing literature and provide more evidence in favor of bundling medical and pharmacy benefits.

As in previous literature, the findings were stronger in chronic condition subpopulations. This result is not surprising given that medical costs in a chronically ill population are higher and more variable than in the general population.^9,10,16 Additionally, this finding makes sense in the context of PHMP as a driver of savings.

We also find evidence that PHMP programs, when combined with integrated pharmacy benefits, generate additional value. We propose that integrated pharmacy benefits result in more timely identification and prioritization of members for clinical outreach from a near real-time feed of both pharmacy and medical claims data. Consistent with this, we find that having integrated benefits increases the likelihood of outreach by health coaches by nearly 11%. Additionally, these data may also better inform nurse health coaches during these outreaches and subsequently make them more effective at managing complex clinical situations or chronic condition profiles. Taken together, offering integrated pharmacy and medical benefits may increase the effectiveness of PHMP.

Although we find cost savings associated with PHMP programs, they do not fully explain all of the savings estimate, with $16 PMPM in savings being associated with members that have integrated medical and pharmacy benefits but do not have a PHMP buy-up. This provides evidence that there may be direct medical cost offsets and that employers who self-fund their benefits should consider these offsets when evaluating pharmacy integration decisions.

Other potential mechanisms may also contribute to the relationship between integrated benefits and reduced cost growth. These include inefficiencies from disjoint benefit design. Multiple aspects of medical and pharmacy benefit design, including utilization management tools like cost sharing, prior authorization policy, step therapy, and formulary inclusion/exclusion, influence both provider incentives and member consumption preferences. Misalignments in these tools can result in a suboptimal consumption mix between pharmacological and medical interventions and ultimately a higher total cost of care.

LIMITATIONS

There are several limitations of this study. We use medical claims data from a single payer in Southeastern Pennsylvania for self-funded employer groups. As such, these findings may not be generalizable to other populations, such as different markets or publicly funded programs. Second, for employers without integrated pharmacy benefits, we do not have access to full pharmacy claims data, including cost. As such, we cannot evaluate differences in spending on pharmacy. Third, it is possible that some companies provide a PHMP benefit to their employees through a third-party vendor, rather than using Independence’s PHMP service. In that scenario, our estimates of the effect of combined integrated pharmacy benefits and PHMP available would be biased toward zero, owing to measurement error in the treatment variable. Lastly, this study analyzes medical costs over a limited time frame, 2022 and 2023, and as such, findings may or may not be generalizable to all years or longer periods.

Conclusions

In this study of commercially insured Independence Blue Cross members, we find that members with integrated medical and pharmacy benefits experienced lower total medical cost growth between 2022 and 2023, compared with members who lacked integrated benefits. This reduction in cost growth was largely driven by lower inpatient and ED spending trends. We also provide evidence that PHMP programs, when combined with integrated pharmacy benefits, generate additional value, potentially owing to more effective use of pharmacy data facilitated by integration. Taken together, these findings indicate that offering integrated medical and pharmacy benefits may yield cost savings and that employers who self-fund their benefits should consider these offsets when evaluating benefit design.

Disclosures

Drs Eric S. Wilkinson, Reetika Kumar, Philip A. Saynisch, and Aaron Smith-McLallen are employees of Independence Blue Cross, a large insurance payer in Southeastern Pennsylvania. Dr Guy David has received consulting fees from Independence Blue Cross.