Association between Medicare annual wellness visits and prevention of falls and fractures in older adults in Texas, USA

Abstract

Fall-related injuries contribute to increased frailty, disability, and premature death in older adults (≥65 years). The US Centers for Medicare and Medicaid Services began reimbursing annual wellness visits (AWVs) in 2011. In the present study, we assessed the effect of AWV receipt in 2017 on fall and fracture prevention through December 31, 2018. Using Texas Medicare data for 2014–2018, we identified cohorts of Medicare beneficiaries ≥68 years, matched for the presence/absence of an AWV in 2017 by propensity score, and observed two outcomes: fracture as a primary diagnosis, and fall occurrences. Rates of each outcome were estimated using the Kaplan–Meier method. Of the 2017 beneficiaries, 32.2% received an AWV. For the 742,494 beneficiaries in the matched cohort, conditional Cox proportional hazards models revealed that receiving an AWV in 2017 was associated with reduced risks for future falls (3.9%) and fractures (4%). The effect of the AWV was stronger on fall reduction in rural residents (HR: 0.799; 95% CI: 0.679 to 0.941) and on fracture reduction in beneficiaries with ≥4 morbidities (HR: 0.918; 95% CI: 0.867 to 0.972). Receipt of an AWV in three consecutive years (2015–2017) further lowered the risk of future falls. We conclude that the risks for future falls/fractures are lower in older adults receiving AWVs. Our study underscores the need for expanded public education programs that raise awareness about AWVs and the potential for AWV data to inform fall prevention interventions and other health promotion practices.

1. Introduction

Injuries from falls are known to contribute to increased frailty, disability, and premature death in community-dwelling older adults ≥65 years of age. Risk factors for falls include environmental conditions, medications, vision, cognition and mood, gait and balance, orthostatic hypotension, osteoporosis, and history of falls.^1–4 Older adults with dementia or cognitive impairment might also have poor postural stability, increasing the risk for falls while walking or transferring.⁵ In 2018, 33.9% of older adults in Texas reported falling at least once, a rate that is higher than the 28% reported for the United States overall.⁶ About 25% of those who fell sustained moderate-to-severe injuries, with some injuries resulting in death, hospitalization, or an emergency department visit.⁶ The rate of injurious falls for community-dwelling older adults has not declined since 2015.^1,2 Fall-related injuries in older adults are potentially preventable with evidence-based fall prevention strategies tailored to the individual’s risks, needs, and preferences.^1,2,7 National policies, public health actions, and clinical practices are urgently needed to ensure that all older adults are engaged in health promotion and disease prevention, including fall prevention.⁸

The American Geriatrics Society (AGS) and the British Geriatrics Society (BGS) have recommended that healthcare providers screen older adults annually for a history of falls or balance impairment.⁹ An algorithm developed by the Centers for Disease Control and Prevention (CDC)¹⁰ implements the clinical practice guidelines for preventing falls published by the AGS and BGS and offers recommendations for risk screening, assessment, and intervention in community-dwelling older adults ≥65 years of age. The algorithm suggests routine fall risk assessment and multifactorial fall prevention interventions that can be provided to individuals who have experienced two or more falls or one fall-related injury.^9–13

The Centers for Medicare and Medicaid Services (CMS) began reimbursing face-to-face annual wellness visits (AWVs) in 2011.^14–16 AWVs are provided free of charge to Medicare beneficiaries, with zero copayments. This preventive service covers brief cognitive impairment assessments, medication reconciliations, and personalized prevention plans such as fall risk assessment and prevention. AWVs could be the avenue for raising awareness in older adults about fall risks, mitigating their personal fall risks through evidence-based fall prevention approaches, and monitoring adherence to their individualized fall prevention plans established during the AWVs. Medicare reimburses fall risk assessment and fall prevention education for clients as part of the AWV.^11–13

To our knowledge, no prior research has examined the mitigating effects of AWVs on the risks for falls and fractures in older adults.¹¹ We therefore set out to explore the effects of AWV services on preventing falls and fractures in community-dwelling older adults. The association of AWV receipt in 2017 with fall and fracture occurrences through December 31, 2018 was assessed. We hypothesized that the risk of experiencing falls and fractures would be lower for Medicare beneficiaries receiving than for those not receiving an AWV in 2017.

2. Methods

2.1. Data source

Using 100% Texas Medicare data, we selected individuals 68 years of age and older who were Medicare beneficiaries from 2014 to 2017 (Table 1; also see supplementary eTable S1 for cohort selection). Beneficiary characteristics and Medicare enrollment status were obtained from the Master Beneficiary Summary File. Outpatient AWVs were determined using the Outpatient Standard Analytic and Carrier files. Comorbidities and history of falls and fractures were obtained from the Medicare Provider Analysis and Review, Outpatient Standard Analytic, and Carrier files. The institutional review board of the University of Texas Medical Branch (Galveston, TX, USA) approved our study.

Table 1.

Characteristics of 1,153,744 Texas Medicare beneficiaries before propensity-score matching, 2017 data

Characteristic	Overall (N)	Annual wellness visit in 2017a				p Valueb
		No		Yes
		(n)	(%)	(n)	(%)
Beneficiaries	1,153,744	782,436	67.8	371,308	32.2
Age group
68–69 Years	99,252	69,396	8.87	29,856	8.04	0.0001
70–74 Years	391,539	264,121	33.76	127,418	34.32
75–79 Years	285,989	189,922	24.27	96,067	25.87
80–84 Years	194,434	130,975	16.74	63,459	17.09
≥85 Years	182,530	128,022	16.36	54,508	14.68
Sex
Male	507,836	354,094	45.26	153,742	41.41	0.0001
Female	645,908	428,342	54.74	217,566	58.59
Race/ethnicity
White	884,274	585,752	74.86	298,522	80.4	0.0001
Black	72,282	52,534	6.71	19,748	5.32
Hispanic	161,380	120,597	15.41	40,783	10.98
Other	35,808	23,553	3.01	12,255	3.3
Original Medicare entitlement
Age	1068,691	718,993	91.89	349,698	94.18	0.0001
ESRD/disabled	85,053	63,443	8.11	21,610	5.82
Dual eligibility
No	1053,641	710,454	90.8	343,187	92.43	0.0001
Yes	100,103	71,982	9.20	28,121	5.82
Morbidities
0	314,460	236,266	30.2	78,194	21.06	0.0001
1	230,161	147,013	18.79	83,148	22.39
2	209,776	135,146	17.27	74,630	20.1
3	145,732	94,640	12.1	51,092	13.76
≥4	253,615	169,371	21.65	84,244	22.69
With ADRDs
No	1051,860	711,626	90.95	340,234	91.63	0.0001
Yes	101,884	70,810	9.05	31,074	8.37
Residential area
Metropolitan	917,841	608,693	77.79	309,148	83.26	0.0001
Nonmetropolitan urban	215,944	158,138	20.21	57,806	15.57
Rural	19,959	15,605	1.99	4,354	1.17

^aDenominator for Beneficiaries percentages: 1,153,744. Denominator for other characteristic percentages (column percentages): in the annual wellness visit No column, 782,436; in the annual wellness visit Yes column, 371,308.

^bBy the chi-squared test for the association between an annual wellness visit (yes/no) and the relevant characteristic. A p value < 0.05 was considered statistically significant.

ESRD = end-stage renal disease; ADRDs = Alzheimer disease and related dementias.

2.2. Cohort selection

The observation window included four Medicare coverage years from January 1, 2014, to December 31, 2017. Receipt of an AWV in 2017 was this study’s event of interest. The study population consisted of community-dwelling older Medicare beneficiaries, 68 years of age or older in 2017, who had been continuously enrolled in both fee-for-service Medicare Part A (insurance for inpatient care and home health care), and Part B (insurance for outpatient care) without Medicare advantage plan covered under Health Maintenance Organizations for the whole years of 2014–2017. Beneficiaries were excluded if they lacked a valid gender identity, were living in long-term nursing homes in 2017, or lacked valid residency information in Texas in 2017 (eTable S1 for cohort selection). Of the 1,153,744 beneficiaries thus included, 32.2% had received an AWV in 2017 (Table 1).

2.3. Measures

Medicare AWVs

A beneficiary covered by the CMS may receive an AWV once every 12 months provided that they have been enrolled in Medicare Part B medical insurance for more than 12 months and have not received an initial preventive physical examination or an AWV providing a personalized prevention plan within the preceding 12 months.

Beneficiaries pay nothing if a healthcare provider accepts the assignment.^14–16 Healthcare providers (i.e., physicians, qualified non-physician practitioners, or medical professionals directly supervised by a physician) can bill G0438 (for an initial AWV) or G0439 (a subsequent AWV) for a beneficiary once each year.^14–16

Identifying fractures and falls

The two outcomes considered in this study were fracture as a primary diagnosis and falls. “Fracture” was defined as an International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), code (eTable S2 all fractures) as the primary diagnosis in either an inpatient, outpatient or carrier claim. “Fall” was defined as any fall (eTable S2 Fall Diagnosis Codes) associated with any position ICD-10 diagnosis code. We created a dichotomous variable to identify fracture as the primary diagnosis and another dichotomous variable to identify the experience of any fall after the index AWV through 12/31/2018.

Covariates

The beneficiary characteristics obtained for the analysis included demographics (age, gender identity, race/ethnicity, and residential area), original Medicare entitlement, dual eligibility, diagnosis of Alzheimer disease or related dementias, receipt of an AWV in 2015 or 2016, falls or fractures that occurred in the 12 months preceding the index AWV, the number of comorbidities in 2016 (using the Elixhauser Comorbidity Index),¹⁷ tobacco use, and osteoporosis in the 2 years preceding the index AWV (using definition from CMS chronic disease data warehous¹⁸). “Residential area” was defined as metropolitan, metropolitan urban, or rural according to the 2013 rural–urban continuum codes from the US Department of Agriculture.¹⁹

2.4. Statistical analyses

To control differences between the beneficiaries with and without an AWV, propensity-score matching was performed. The propensity score of receipt of an AWV was generated using a logistic regression model including age, gender identity, race/ethnicity, original entitlement, dual eligibility, number of comorbidities, Alzheimer disease and related dementias, and residential area.

For each beneficiary with an AWV, we performed greedy nearest neighbor matching to select one beneficiary without an AWV within a caliper equal to 0.2 standard deviations of the logit of the propensity score. The date of the AWV for the beneficiary receiving the AWV was assigned to the matched non-AWV beneficiate. Then, pairs in which the non-AWV beneficiary’s assigned index date fell after death were excluded. This resulted in 371,247 propensity-score matched pairs.

The frequencies of covariates for beneficiaries with and without an AWV were calculated before and after propensity-score matching and were compared using the chi-squared test. After propensity-score matching, standardized differences were also calculated to assess the balance of covariates between the groups. Separate Kaplan–Meier failure curves were generated for any falls and primary diagnoses of fractures by AWV status. Beneficiaries were censored at lost coverage, death, or study end (12/31/2018). The impacts of AWV receipt in 2017 on the two outcomes of interest were estimated as hazard ratios (HRs) obtained from a conditional Cox proportional hazards model adjusted for the specific study outcome in the year before the index date.

We tested the interaction between each covariate and the AWV in the conditional Cox proportional hazards models for each outcome. In addition, we tested the interaction between previous receipt of AWVs in 2015 and 2016, and receipt of an AWV in 2017. To compare beneficiaries receiving no AWVs during 2015–2017, we removed 73,897 pairs in which non-AWV 2017 beneficiaries received an AWV during 2015–2016. The remaining 297,350 pairs were entered into the sensitivity analyses examining interaction effects of the AWV in 2017 and the AWVs in 2015 and 2016 for falls and fractures. All analyses were performed in the SAS software application (version 9.4: SAS Institute, Cary, NC, USA).

3. Results

3.1. Characteristics of the Medicare fee-for-service beneficiaries

As shown in Table 1, beneficiaries receiving an AWV were younger; more likely to be female, White, eligible for Medicaid coverage, resident in a metropolitan area; also more likely to have more comorbidities and a Medicare original entitlement of Aging; and to be less likely to have Alzheimer disease and related dementias. Table 2 shows that the characteristics used to generate propensity scores were balanced in the groups, with a standardized difference of less than 10%. The AWV group was less likely than the non-AWV group to have experienced falls and fractures in the year before the index date. About 20% of beneficiaries in the non-AWV group, compared with 61.3% in the AWV group, had received an AWV during 2015–2016.

Table 2.

Characteristics of 742,494 Texas Medicare beneficiaries^a after propensity-score matching, 2017 data

Characteristic	Overall (N)	Annual wellness visit in 2017				p Valueb	Std. diff.c
		No		Yes
		(n)	(%)	(n)	(%)
Beneficiaries	742,494	371,247	50	371,247	50
Age group
68–69 Years	59,709	29,854	8.04	29,855	8.04	0.9991	0.0005
70–74 Years	254,838	127,434	34.33	127,404	34.32
75–79 Years	192,123	96,063	25.88	96,060	25.87
80–84 Years	126,940	63,495	17.1	63,445	17.09
≥85 Years	108,884	54,401	14.65	54,483	14.68
Sex
Male	307,422	153,711	41.41	153,711	41.41	1	0
Female	435,072	217,536	58.6	217,536	58.6
Race/ethnicity
White	597,263	298,791	80.48	298,472	80.4	0.5597	0
Black	39,368	19,622	5.29	19,746	5.32
Hispanic	81,548	40,772	10.98	40,776	10.98
Other	24,315	12,062	3.25	12,253	3.3
Original Medicare entitlement
Age	693,687	344,048	92.67	349,639	94.18	0.0001	−0.0608
ESRD/disabled	48,807	27,199	7.33	21,608	5.82
Dual eligibility
No	686,451	343,320	92.48	343,131	92.43	0.4064	0.0019
Yes	56,043	27,927	7.52	28,116	7.57
Morbidities
0	156,535	78,344	21.1	78,191	21.06	0.995	0
1	166,237	83,097	22.38	83,140	22.39
2	149,225	74,607	20.1	74,618	20.1
3	102,104	51,021	13.74	51,083	13.76
≥4	168,393	84,178	22.67	84,215	22.68
With ADRDs
No	680,571	340,386	91.69	340,185	91.63	0.3989	0.002
Yes	61,923	30,861	8.31	31,062	8.37
Residential area
Metropolitan	618,175	309,078	83.25	309,097	83.26	0.9833	0
Nonmetropolitan urban	115,594	57,798	15.57	57,796	15.57
Rural	8,725	4,371	1.18	4,354	1.17
Tobacco used
No	700,254	349,124	94.04	351,130	94.58	0.0001	−0.0233
Yes	42,240	22,123	5.96	20,117	5.42
Osteoporosise
No	683,485	345,616	93.1	337,869	91.01	0.0001	0.0772
Yes	59,009	25,631	6.9	33,378	8.99
Previous fallf
No	684,126	341,041	91.86	343,085	92.41	0.0001	−0.0205
Yes	58,368	30,206	8.14	28,162	7.59
Previous fracturef
No	729,348	364,158	98.09	365,190	98.37	0.0001	−0.0211
Yes	13,146	7,089	1.91	6,057	1.63
Previous AWV
Only in 2015	63,735	20,525	5.53	43,210	11.64	0.0001	0.9718
Only in 2016	117,572	33,208	8.94	84,364	22.72
In both 2015 and 2016	120,254	20,164	5.43	100,090	26.96
None	440,933	297,350	80.09	143,583	38.68

^aExcludes 61 pairs with an event record after the recorded death date.

^bBy the chi-squared test. A p value < 0.05 was considered statistically significant (the variable was significantly associated with case/control [i.e., annual wellness visits: 1/0]).

^cThe propensity-score matching considered gender identity, ADRDs, dual eligibility, age, race/ethnicity, residence, and comorbidities. Because all standard differences for the covariates were <0.1, the propensity-score matching was accepted as well balanced.

^dFor tobacco use, we included valid the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10)/CPT4/HCPCS codes3: DX F17.200, F17.201, F17.203, F17.208, F17.209, F17.210, F17.211, F17.213, F17.218, F17.219, F17.220, F17.221, F17.223, F17.228, F17.229, F17.290, F17.291, F17.293, F17.298, F17.299, O99.330, O99.331, O99.332, O99.333, O99.334, O99.335, T65.211A, T65.212A, T65.213A, T65.214A, T65.221A, T65.222A, T65.223A, T65.224A, T65.291A, T65.292A, T65.293A, T65.294A, Z72.0 (any DX on the claim); HCPCS codes: 99406, 99407, G9276, G9458; valid the International Statistical Classification of Diseases and Related Health Problems, 9th Revision (ICD-9)/MS DRG/HCPCS codes3: DX 305.1, 649.00, 649.01, 649.02, 649.03, 649.04, 989.84 (any DX on the claim); and HCPCS Codes: 99406, 99407.

^eFor osteoporosis, we included ICD-10 codes: M81.0, M81.6, M81.8.

^fWithin the 12 months preceding the index date.

ESRD = end-stage renal disease; ADRDs = Alzheimer disease and related dementias; AWV = annual wellness visit.

3.2. Associations between AWV receipt and falls

As shown in Figure 1a, the fall rates at 24 months of follow-up for the AWV and non-AWV groups from the Kaplan–Meier curves were 16.61% and 17.31% respectively. AWV receipt in 2017 was associated with a 3.9% reduction in the risk for falls (HR: 0.961; 95% confidence interval [CI]: 0.948 to 0.974) with adjustment for a previous fall (eTable S3, Models 1). Further control based on AWV receipt during 2015–2016, tobacco use, and osteoporosis, slightly attenuated the effect of AWV receipt in 2017 on falls (HR: 0.974; 95% CI: 0.958 to 0.991) (eTable S3, Models 2).

Figure 1.

Kaplan–Meier failure graphs with customized y-axes (failure probability) for (a) falls and (b) fractures as a primary diagnosis. The p value of the log-rank test in both cases is significant at the 0.05 level. AWV = annual wellness visit.

Table 3, Model A, shows stratified results for the effect of 2017 AWV receipt on risk of fall at each level of a covariate for which a significant interaction between that covariate and AWV receipt was found in the conditional Cox proportional hazards models. The effect of 2017 AWV receipt on fall reduction was stronger for those receiving an AWV in both 2015 and 2016 or those receiving an AWV in 2016 (respectively, HR: 0.887; 95% CI: 0.799 to 0.985; HR: 0.899; 95% CI: 0.819 to 0.987), and rural residents (HR: 0.799; 95% CI: 0.679 to 0.941).

Table 3.

Cox model results for 742,494 propensity-score matched Medicare beneficiaries with or without an annual wellness visit (AWV) in 2017

Model	Parameter	Comparator	HR	95% CI
A. Significant interaction effects for 2017 AWV receipt in predicting falls
	Previous AWV	AWV in both 2015 and 2016	0.887	0.799–0.985
		No AWV in either 2015 or 2016	0.992	0.967–1.017
		AWV in 2015 only	0.933	0.809–1.077
		AWV in 2016 only	0.899	0.819–0.987
	Comorbidities	0	1.073	1.022–1.126
		1	0.943	0.907–0.981
		2	0.97	0.934–1.008
		3	0.947	0.908–0.988
		≥4	0.969	0.942–0.996
	Race/ethnicity	White	0.971	0.953–0.989
		Black	0.935	0.862–1.015
		Hispanic	1.029	0.981–1.079
		Other	0.858	0.766–0.961
	Residence	Metropolitan	0.974	0.956–0.992
		Nonmetropolitan urban	0.983	0.942–1.025
		Rural	0.799	0.679–0.941
B. Significant interaction effects for 2017 AWV receipt in predicting fracture as the primary diagnosis
	Comorbidities	0	1.081	0.982–1.189
		1	0.968	0.893–1.048
		2	1.008	0.932–1.09
		3	1.006	0.923–1.097
		≥4	0.918	0.867–0.972

HR = hazard ratio; CI = confidence interval.

3.3. Associations AWV receipt and fractures

As shown in Figure 1b, the fracture rates at 24 months of follow-up for the AWV and non-AWV groups from the Kaplan–Meier curves were 3.87% and 3.98% respectively. AWV receipt in 2017 was associated with a 4% reduction in the risk for facture (HR: 0.960; 95% CI: 0.934 to 0.988) with adjustment for a previous facture. However, with further control based on AWV receipt during 2015–2016, tobacco use, and osteoporosis, that association became insignificant (HR: 0.978; 95% CI: 0.946 to 1.102; eTable S3, Models 3 and 4).

Table 3, Model B, shows stratified results for the effect of 2017 AWV receipt on risk on facture at each level of a covariate for which a significant interaction between that covariate and AWV receipt was found in the conditional Cox proportional hazards models. The effect of 2017 AWV receipt on fracture reduction was stronger for beneficiaries with ≥4 comorbidities (HR: 0.918; 95% CI: 0.867 to 0.972).

3.4. Sensitivity analyses

Table 4 shows comparing to beneficiaries without AWV in 2015–2017, the effect of 2017 AWV receipt on reduction in falls or fractures after adjusted for previous falls or previous fractures, tobacco use and osteoporosis. The effect was stronger for beneficiaries with an AWV in 2015 and 2016 (fall: HR: 0.898; 95% CI: 0.871 to 0.926; fracture: HR: 0.896; 95% CI: 0.842 to 0.954). The effect was also stronger for beneficiaries receiving an AWV in 2016 only (fall: HR: 0.944; 95% CI: 0.913 to 0.975; facture: HR: 0.905; 95% CI: 0.846 to 0.968).

Table 4.

Sensitivity analyses for 297,350 Medicare beneficiary pairs (594,700 observations) with and without an annual wellness visit (AWV) in 2017^a

Analysis	Parameter	Comparator	HR	95% CI
1. Predicting falls in 2017 and 2018: significant interaction effects for receiving an AWV in 2017 and receiving a previous AWV
	Previous AWV	AWV in both 2015 and 2016	0.898	0.871–0.926
		No AWV in either 2015 or 2016	0.992	0.967–1.017
		AWV in 2015 only	1.0	0.958–1.044
		AWV in 2016 only	0.944	0.913–0.975
2. Predicting fracture as the primary diagnosis in 2017 and 2018: significant interaction effects for receiving an AWV in 2017 and receiving a previous AWV
	Previous AWV	AWV in both 2015 and 2016	0.896	0.842–0.954
		No AWV in either 2015 or 2016	1.004	0.953–1.057
		AWV in 2015 only	1.005	0.92–1.099
		AWV in 2016 only	0.905	0.846–0.968

^aExcludes 73,897 pairs (147,794 observations) in which the control beneficiary received an AWV in either or both of 2015 and 2016.

HR = hazard ratio; CI = confidence interval.

3.5. Additional sensitivity analyses

To ensure our study results were robust for the definition of fall and fracture, we used CQ3 algorithm from Mintz et al. paper-the outpatient and provider claims with fall²⁰ to identify fall, and searched for any diagnosis positions to identify fracture. The difference of fall and fracture between patients with and without AWV from these new definitions were quite similar to our main analyses (eTable S4, eTable S5) suggesting that the codes in eTable S2 produced robust results.

4. Discussion

4.1. Discussion and practical implications

We observed that, when community-dwelling older adults received an AWV in 2017, their risk of experiencing falls and fractures over the subsequent 24 months was about 4% lower than the risk for those who did not receive an AWV in 2017. The study findings thus support our hypothesis. The findings of this study align with the AGS and BGS recommendations that healthcare providers should screen adults ≥65 years of age annually for a history of falls.⁹ The US CDC encourage healthcare providers to use the AGS and BGS clinical practice guidelines for preventing falls and fall-related fractures, and provide recommendations for fall risk screening, assessment, and interventions for older adults ≥65 years of age.^9–13

AWVs would be ideal for delivering fall prevention patient education and introducing interventions relevant to the risk factors faced by older adults (e.g., removing or replacing potentially inappropriate medications^21,22). In our study, it shows that two or three consecutive years AWV receipt associated with a higher decreased risk of future falls and fractures. That finding highlights the value of assessing and addressing changes in risk factors annually and “nudging” community-dwelling older adults toward fall prevention self-care.

The effect size of the 2017 AWV on fall reduction was larger for rural residents, who experienced a 20.1% lower risk of falling. That observation further affirmed that rural and urban community-dwelling older adults might have different risk factors for falls. Intuitively, hospitals and medical clinics might be assumed to be more accessible for urban residents than rural residents. Our study findings did not support that assumption, however. Our result implies that, compared with urban residents, rural residents might have a more socially active and physically outdoor-focused lifestyle²³ with a higher risk of falls; therefore, AWV was associated with a large effect on fall reduction. The wellbeing of older adults might link less to the density of medical care services in their residential areas, and more to their chosen lifestyle and wellness practice. Future research using Medicare beneficiary files and other health- and wellbeing-related databases (e.g., U.S. census data) is needed to verify this hypothesis.

Critical access hospitals and medical clinics in rural areas could have played a key role in promoting AWV delivery for rural residents. However, Medicare data do not capture how AWVs are marketed and implemented in rural areas versus metropolitan and nonmetropolitan urban areas. An important area for future study is a comprehensive examination of the campaign strategies that local governments in rural areas adopt to promote preventive health services and urge AWV delivery by prioritizing AWV receipt rates. Increasing AWV delivery could potentially be achieved by operating mobile clinics or stationing clinics at community events or harvest-related festivals in rural areas. We speculate that trust-building between health systems, local governments, and rural residents, and meeting older adults where they are (e.g., fall prevention strategies tailored to the individual’s risks, needs, and preferences^1,2,7) could have contributed to the significant benefit of AWV receipt in lowering the risk of falls. Future research to identify effective implementation strategies for AWVs is needed.

The effect on fracture reduction of AWV receipt in 2017 was stronger for older adults with ≥4 comorbidities, who had an 8.2% lower risk of future fractures. Such individuals might have derived greater benefit from the AWV-required medication reconciliation of the multiple medications prescribed for their multiple comorbidities. Medication reconciliation—the process of comparing a patient’s medical conditions with all the prescribed and over-the-counter medications that the patient has been taking—intends to avoid medication errors resulting from omissions, duplications, dosing errors, or drug interactions.^21,22 Additional research is warranted to confirm the benefit of person-centered AWVs related to tailored interventions for fall and fracture prevention.

4.2. Study limitations

Our study has some limitations. The analyses were limited to Medicare fee-for-service beneficiaries in Texas and to Medicare billing data, which do not capture how healthcare providers perform AWVs (e.g., actively engaging beneficiaries alone vs. engaging beneficiaries and their family caregivers together, using strategies to engage beneficiaries in self-care related to fall and fracture prevention). As a result, we could not determine whether appropriate discussions were held with the beneficiaries (e.g., whether the priorities and health goals of the beneficiaries to prevent falls and fractures were discussed).

Our observed association might possibly be attributed, at least in part, to residual confounding, in that beneficiaries with healthy behaviors would be more likely to have an AWV. We attempted to address that issue by using propensity-score matching. We also adjusted the conditional Cox proportional hazard regression model analysis for the previous AWVs, previous falls or fractures, tobacco use, and osteoporosis. Future study controlling for other health behaviors, such as physical activity from survey and Medicare linkage data, is needed. Another limitation is that potentially inappropriate medications^21,22 before AWV, which could be associated with a higher risk of fractures, were not included as covariates in the analyses. Potentially inappropriate medication use could also reduce after medication reconciliation during AWV, which could mediate the association between AWV and fall/fracture. This topic deserves a separate study for Medicare enrollees with continuous enrollment in the Part D drug plan. Our study also did not capture minor falls not requiring medical services. In addition, information about the physical environment at home (e.g., presence or absence of ramps or railings, bathroom grab bars, and night lights) that might affect the study outcome was unavailable.

4.3. Future research

Future research using electronic health records and qualitative methods are needed to understand how AWVs are delivered, especially with respect to revealing the most effective strategies for educating older adults in self-care to prevent falls and fractures in various AWV delivery settings (e.g., health system–affiliated clinics, independent practices). Data from such studies can guide development and implementation of AWV protocols based in the strongest evidence for reduction of falls and fractures in community-dwelling older adults. An important focus would be a comprehensive examination of the strategies that local governments in rural areas have successfully adopted to promote preventive health services and fall prevention triggered by AWV receipt. Data from such a study could inform other communities about how to expand AWV delivery by all clinicians caring for the Medicare population.

5. Conclusions

During a 24-month follow-up period, older adults in Texas were less likely to experience falls and fractures when they had received a Medicare AWV than when they had not. Using the AWV to deliver fall and fracture education could be one strategy to address the current fall and fracture epidemic in Texas.⁶ Falls and fall-related injuries are potentially preventable in community-dwelling older adults, and thus awareness and use of Medicare AWVs in day-to-day clinical practice should be increased.^1,2,7 National policies, public health actions, and clinical practices are vital to ensuring that all older adults are engaged in self-care for fall and fracture prevention.⁸

The US CMS could use public campaigns targeted at community-dwelling older adults and their family caregivers to raise awareness about the preventive medicine goals of AWVs. Such campaigns should focus on AWVs being free of charge to Medicare beneficiaries. The CMS could also potentially introduce incentives for healthcare providers (e.g., one of the new quality measures linked to Medicare payments) to deliver AWVs annually to every older adult who is eligible for Medicare Part B preventive services, which cover services to prevent or detect illness. Beneficiaries with Medicare Part B pay nothing for most preventive services. For example, to promote safety and prevent falls and fractures in older adults living in our communities, a “Have you had your free annual wellness visit this year?” slogan targeted to places where such individuals gather could be the gentle nudge needed. Development of such slogans should always consider local culture and customs (e.g., adding wellbeing-related Bible verses for faith-based communities, or role models for ethnic communities).

HIGHLIGHTS.

• Older adults having an annual wellness visits (AWV) had less fall and fracture risks.

• Receiving an AWV was associated with reduced risks for future falls and fractures.

• The effect of AWV was stronger on fall reduction in rural residents.

• The AWV effect was stronger on fracture reduction in those with ≥4 morbidities.

• Receipt of an AWV in 3 consecutive years further lowered the risk of future falls.

Data statement

The data used in this manuscript is a research identified file from the Centers for Medicare and Medicaid Services, which do not permit such submissions.