Abstract
Background:
Limited access to primary care may disproportionately affect older adults, who often have greater chronic disease management and care coordination needs. However, little is known about the effect of having a primary care practitioner (PCP) on longevity in the aging population.
Objective:
To examine the association of having a usual source of primary care with mortality and life expectancy among US adults aged 65 and older.
Design:
Retrospective cohort study, using nationally representative data from the 2000 and 2005 cohorts of the National Health Interview Survey linked with National Death Index records through 2019.
Participants:
All respondents aged 65 to 84 (n=10,873, weighted n=16,484,914)
Interventions/Exposures:
Having a usual source of primary care
Main Measures:
Using a Cox proportional hazards model, we examined the association between exposure to primary care and 15-year mortality, adjusting for sociodemographic factors and respondent life expectancy (using a validated index). We also used this model to generate survival curves by exposure to primary care and computed median survival times for each group.
Key Results:
Overall, 60.3% of respondents were female, 83.6% were non-Hispanic White, and 6.4% (n=739, weighted n=1,056,554) did not have a usual source of primary care. Use of primary care was associated with a lower 15-year mortality risk (aHR: 0.84, 95% CI: 0.72–0.98). Median survival time was also at least 2.1 years longer among those who used primary care (>15 years) compared to those who did not (12.9 years).
Conclusions:
We found that primary care use is associated with greater survival among older adults. As the population of adults aged 65+ is growing rapidly, investing in primary care is essential for the health of US older adults.
Keywords: Primary Care, Mortality, Life Expectancy
INTRODUCTION
Access to primary care has become increasingly strained in the United States (US) and other high-income countries due to workforce shortages, income disparities between primary care practitioners (PCPs) and specialists, geographic maldistribution of PCPs, and patient socioeconomic factors such as poverty and lack of health insurance.[1] Seventy-five million Americans live in a primary care Health Professional Shortage Area (HPSA) and over 100 million—almost a third of the US population—lack access to a usual source of primary care.[2,3] These shortages are of particular concern for adults aged 65 and older, who are projected to make up 22% of the US population by 2040.[4] Older adults often have greater primary care-related needs due to increased numbers of comorbidities, prescription medications, and specialist visits, requiring extensive efforts to manage chronic disease and coordinate care.[5] Given the greater use of acute care among older adults, access to primary care is also crucial for preventing unnecessary hospitalizations and readmissions through early detection and intervention, as well as effective after-care transitions.[6,7] As the population of adults aged 65 and older increases, the shortage of PCPs is expected to grow more than sixfold.[2,4] This issue is particularly pronounced among geriatrician PCPs and consultants, as two thirds of US counties lack any geriatrician physician or nurse practitioner whatsoever.[8]
Despite these well-documented shortages, little is known about the effect of having a PCP on longevity, especially among older adults, which could guide future investments in primary care. A substantial body of literature has found that primary care improves health outcomes among the overall population, with more pronounced benefits for low-income and disadvantaged patient groups.[9–12] For example, studies have found increased PCP supply to be associated with increased life expectancy and decreased mortality at the county level.[13–15] By contrast, primary care shortages have been linked with increased population mortality due to delayed detection and treatment of diseases, reduced uptake of preventive services, and worse chronic disease management leading to higher rates of complications and death.[13–17]
It is estimated that increasing the PCP supply to the minimum threshold of 1 PCP per 3500 people in shortage areas could increase the mean life expectancy by 22.4 days for the entire county, or 56.3 days if the ratio is increased to 1 PCP per 1500 people.[13] However prior ecological studies do not reveal the effects of primary care on individual mortality and do not provide disaggregated data about the effects on specific high-risk groups, such as older adults. Evaluating this individual-level association is essential given the rapidly aging population, which faces a disproportionate burden of complex health needs requiring ongoing management and care coordination.[5] Understanding the survival benefit of primary care could also inform policy for health system design and adaptation, and guide resource allocation to determine the most effective ways to improve survival among this vulnerable population.
In this study, we aimed to examine how having a usual source of primary care influenced mortality among US adults aged 65 and older. We hypothesized that receipt of primary care would be associated with increased longevity after adjusting for other factors associated with mortality.
METHODS
We used data from the 2000 and 2005 cohorts of the National Health Interview Survey (NHIS), an annual nationally representative cross-sectional survey of the US non-institutionalized civilian population. Administered by the National Center for Health Statistics, the NHIS collects information about individuals’ health and utilization of medical services. Data on participant mortality is ascertained through a probabilistic match between NHIS and National Death Index (NDI) records, with current follow-up through 2019. Response rates for adults were 72.1% in 2000 and 69.0% in 2005. We included all individuals aged 65 to 84 (n=10,873, weighted n=16,484,914); individuals aged 85 and older were excluded as the NHIS does not collect disaggregated age data for this population.
Our independent variable was having a usual source of primary care, which we defined as (i) having a usual place of care for “routine or preventive care, such as a physical exam or check up” that was (ii) a “clinic or health center” or a “doctor’s office or HMO [health maintenance organization, a health insurance group that provides services for a fixed annual fee]” (in contrast to other locations like a “hospital emergency room”). This definition of primary care (having a usual place of care that provides routine or preventive care) has been used in prior studies and captures the heterogeneity of primary care in the US.[18–20] In several countries, patients must see their PCP before being referred to specialty services.[21] In the US, many patients may see a specialist without seeing a PCP first. Additionally, primary care is often provided by a diverse range of specialties (e.g., family medicine, general internal medicine, general pediatrics, geriatrics) and practiced with a team-based approach where patients may be assigned to a physician, but see an allied health professional (e.g., physician assistant, nurse practitioner) as issues arise. Our primary outcomes were 15-year mortality and median survival time.
Statistical Analyses
Weighted prevalence and 95% confidence intervals (CIs) were calculated using survey procedures in SAS to reflect the NHIS study design’s representative sampling. Subgroup analyses were conducted by 5-year age groups (65–69, 70–74, 75–79, 80–84). Mortality status was determined through linkage with NDI data available through 2019.
First, we used a Cox proportional hazards model with survey weighting to examine the association between exposure to primary care and death within 15 years. Then, we examined the effect of primary care on mortality adjusting for health and sociodemographic factors. To account for health in our model, we used the validated Schonberg mortality index. This index was developed for adults aged 65 and older using NHIS data from 1997 to 2000 with follow-up mortality data through 2011.[22–24] The 11-item index takes into account multiple health factors (including age, sex, body mass index, smoking, history of diabetes, emphysema, or cancer, hospitalizations in the past year, self-rated health, mobility, and functional limitations). The index assigns points to each risk factor present based on the original survival model’s beta coefficients. For each participant, we calculated a mortality risk score using this index. Past studies have found the index to have excellent calibration and discrimination (c-index 0.72) in predicting mortality up to 14 years.[24] We included risk scores in the model as a continuous variable.
Based on a literature review, we also adjusted for sociodemographic factors known to be associated with all-cause mortality that were not included in the Schonberg mortality index including race/ethnicity, educational attainment, family income, health insurance, and region.[25–29] These were modeled as categorical variables (categories and reference groups indicated in Table 3). Our final model incorporated five imputed income datasets provided by NHIS to address item nonresponse for family income in accordance with National Center for Health Statistics protocols.[30–31] Rates of missing data were low for the other variables, which ranged from 0.1% to 1.4% across variables.
Table 3:
Adjusted Cox proportional hazards model predicting 15-year mortality in 2000 and 2005 combined NHIS cohorts
| n overall=10,245 | n deaths= 3379 | |||
|---|---|---|---|---|
| Variable | Category | Hazard Ratio | 95% CI | p value |
| Exposure to Primary Care | No | - | - | - |
| Yes | 0.84 | (0.72–0.98) | 0.03 | |
| Race/ethnicity | Non-Hispanic White (reference) | - | - | - |
| Asian | 0.60 | (0.45–0.80) | <0.001 | |
| Black | 0.80 | (0.70–0.92) | 0.001 | |
| Hispanic | 0.75 | (0.63–0.89) | 0.001 | |
| Other | 0.76 | (0.34–1.70) | 0.51 | |
| Education | Less than high school | 1.14 | (1.03–1.27) | 0.02 |
| High school/GED | 1.00 | (0.90–1.10) | 0.93 | |
| Some college (reference) | - | - | - | |
| Bachelor’s or higher | 0.92 | (0.80–1.05) | 0.22 | |
| Health Insurance | Private or Medicare (reference) | - | - | - |
| Medicaid | 0.92 | (0.79–1.07) | 0.28 | |
| Uninsured | 0.89 | (0.54–1.47) | 0.66 | |
| Other | 1.21 | (0.80–1.85) | 0.36 | |
| Region | Northeast (reference) | - | - | - |
| Midwest | 1.04 | (0.93–1.16) | 0.54 | |
| South | 0.97 | (0.87–1.08) | 0.59 | |
| West | 1.03 | (0.91–1.16) | 0.67 | |
| Family income | < $15,000 | 1.69 | (1.41–2.02) | <0.001 |
| $15,000–$24,999 | 1.52 | (1.27–1.81) | <0.001 | |
| $25,000–$44,999 | 1.31 | (1.09–1.57) | 0.007 | |
| $45,000–$74,999 | 1.13 | (0.93–1.37) | 0.25 | |
| $75000 and over (reference) | - | - | - | |
| Schonberg indexa (per point increase) | 1.09 | (1.08–1.10) | <0.001 |
Using this model, we generated survival curves by primary care exposure and computed median survival times for each group. We defined median survival as the time at which the estimated survival probability was equal to 0.5. To compute median survival times by age group, we used a secondary model adding age as a covariate. Models were run as complete case analyses.
Finally, to better understand mediators of the relationship between primary care and mortality, we examined the effect of adding receipt of flu vaccine, pneumonia vaccine, vigorous exercise, moderate exercise, and colorectal cancer screening to our model, individually and combined.[32–35] We chose to model the effect of colorectal cancer screening as representative of other cancer screening tests since both men and women are eligible for this screening. When examining the mediating effect of colorectal cancer screening, we limited our sample to adults aged 65–75.
All analyses were performed using SAS version 9.4. All tests were 2-sided; P-values<0.05 and 95% CIs that did not cross 1 defined statistical significance. As we used a public, de-identified dataset, this study did not meet the criteria for human subject research and did not require institutional review board approval.
RESULTS
Overall, 60.3% of the 2000 and 2005 NHIS participants were female, 83.6% were non-Hispanic White, and 6.4% (739, weighted n=1,056,554) did not have a usual source of primary care. Those who reported use of primary care were more likely to be older, female, and non-Hispanic White and have higher educational attainment, family income, and private health insurance. The two exposure groups differed across health-related factors in the Schonberg index; those exposed to primary care were more likely to have a history of diabetes or cancer, and to have never smoked (Table 1).
Table 1:
Sociodemographic and health characteristics by exposure to primary care among 2000 and 2005 NHIS participants
| Overall n=10,873 weighted n = 16,484,914 | Exposed to primary care n=10,134 weighted n = 15,428,360 | Not exposed to primary care n=739 weighted n = 1,056,554 | ||
|---|---|---|---|---|
| weighted % (95% CI) | weighted % (95% CI) | weighted % (95% CI) | ||
| Survey year | 2000 | 49.3 (48.2–50.3) | 49.1 (48.0–50.1) | 52.0 (48.1–55.8) |
| 2005 | 50.7 (49.7–51.8) | 50.9 (49.9–52.0) | 48.0 (44.2–51.9) | |
| Age | 65–69 | 29.1 (28.1–30.0) | 28.8 (27.8–29.7) | 33.6 (29.8–37.4) |
| 70–74 | 27.7 (26.8–28.6) | 27.6 (26.6–28.6) | 29.1 (25.7–32.6) | |
| 75–79 | 24.6 (23.7–25.5) | 24.8 (23.9–25.7) | 22.3 (19.1–25.6) | |
| 80–84 | 18.6 (17.7–19.5) | 18.9 (17.9–19.8) | 15.0 (12.0–17.9) | |
| Sex | Male | 39.7 (38.6–40.7) | 38.5 (37.5–39.6) | 56.2 (52.2–60.1) |
| Female | 60.3 (59.3–61.4) | 61.5 (60.4–62.5) | 43.8 (39.9–47.8) | |
| Exposure to primary care | Yes | 93.6 (93.1–94.1) | 100 | 0 |
| No | 6.4 (5.9–6.9) | 0 | 100 | |
| Race/ethnicity | Non-Hispanic White | 83.6 (82.5–84.7) | 84.3 (83.2–85.3) | 74.2 (70.6–77.9) |
| Hispanic | 5.3 (4.7–5.9) | 5.1 (4.6–5.6) | 8.4 (5.8–11.1) | |
| Black | 8.9 (8.1–9.6) | 8.5 (7.7–9.3) | 14.1 (11.6–16.5) | |
| Asian | 1.8 (1.5–2.2) | 1.8 (1.4–2.2) | 2.6 (1.3–3.8) | |
| Other | 0.3 (0.2–0.4) | 0.3 (0.2–0.4) | 0.5 (0.0–1.2) | |
| Unknown | 0.1 (0.03–0.14) | 0.1 (0.02–0.14) | 0.2 (0.0–0.5) | |
| Education | Less than high school | 27.4 (26.3–28.6) | 27.0 (25.8–28.1) | 34.1 (30.5–37.7) |
| High School | 34.1 (32.9–35.2) | 34.3 (33.1–35.5) | 30.2 (26.5–34.0) | |
| Some college | 20.1 (19.3–20.9) | 20.3 (19.4–21.1) | 18.1 (15.2–21.1) | |
| Bachelor’s or higher | 17.0 (16.0–18.0) | 17.2 (16.2–18.2) | 14.1 (11.4–16.9) | |
| Unknown | 1.4 (1.1–1.6) | 1.2 (1.0–1.5) | 3.4 (2.0–4.8) | |
| Health Insurance | Private/Medicare | 91.0 (90.4–91.6) | 91.8 (91.2–92.4) | 79.9 (76.6–83.3) |
| Medicaid | 7.5 (6.9–8.1) | 7.3 (6.6–7.8) | 11.8 (9.0–14.7) | |
| Uninsured | 0.7 (0.6–0.9) | 0.5 (0.3–0.6) | 4.1 (2.7–5.6) | |
| Other | 0.8 (0.6–0.9) | 0.6 (0.4–0.7) | 4.1 (2.7–5.5) | |
| Region | Northeast | 21.0 (19.8–22.1) | 21.0 (19.9–22.1) | 19.8 (16.5–23.2) |
| Midwest | 24.3 (23.1–25.5) | 24.3 (23.1–25.6) | 23.3 (20.0–26.6) | |
| South | 36.6 (35.2–37.9) | 36.6 (35.2–38.0) | 36.3 (32.5–40.0) | |
| West | 18.2 (16.9–19.5) | 18.0 (16.8–19.3) | 20.6 (16.8–24.3) | |
| Family income | < $15,000 | 28.2 (27.1–29.3) | 27.4 (26.3–28.5) | 40.0 (36.2–43.8) |
| $15,000–$24,999 | 24.7 (23.9–25.5) | 24.8 (24.0–25.6) | 23.3 (20.9–25.8) | |
| $25,000–$44,999 | 26.2 (25.4–27.1) | 26.6 (25.7–27.5) | 21.0 (18.2–23.9) | |
| $45,000–$74,999 | 12.2 (11.6–12.9) | 12.4 (11.7–13.0) | 10.2 (8.1–12.4) | |
| $75000 and over | 8.6 (8.0–9.3) | 8.9 (8.2–9.5) | 5.4 (3.6–7.2) | |
| Schonberg indexa mean (95% CI) [range] | 8.0 (7.9–8.1) [0–25] | 8.0 (7.9–8.1) [0–25] | 7.1 (7.9–8.4) [0–23] | |
| The variables below are included in the Schonberg index. They were not separately modeled. | ||||
| Body mass index | <25 | 38.3 (37.4–39.3) | 38.2 (37.3–39.2) | 39.4 (35.5–43.4) |
| ≥25 | 61.7 (60.7–62.6) | 61.8 (60.8–62.7) | 60.6 (56.6–64.5) | |
| Perceived health | Excellent/very good | 40.7 (39.7–41.6) | 40.6 (39.6–41.6) | 42.3 (38.6–46.1) |
| Good | 34.5 (33.5–35.5) | 34.8 (33.8–35.7) | 31.1 (27.5–34.7) | |
| Fair/poor | 24.7 (23.8–25.6) | 24.6 (23.7–25.5) | 26.2 (22.8–29.6) | |
| Unknown | 0.08 (0.03–0.13) | 0.05 (0.01–0.1) | 0.4 (0.0–0.8) | |
| Emphysema/chronic bronchitis | 10.1 (9.5–10.7) | 10.1 (9.5–10.7) | 10.6 (8.1–13.1) | |
| Diabetes | 16.0 (15.3–16.7) | 16.3 (15.6–17.1) | 10.9 (8.5–13.2) | |
| Cancer | 20.3 (19.4–21.1) | 20.6 (19.7–21.5) | 15.3 (12.4–18.2) | |
| Dependent in at least one IADLb | 10.8 (10.2–11.4) | 10.8 (10.2–11.4) | 11.0 (8.5–13.4) | |
| Difficulty walking several blocks | Not at all difficult | 58.9 (58.0–59.9) | 58.7 (57.7–59.6) | 62.2 (58.4–65.9) |
| A little difficult to very difficult | 40.8 (39.8–41.7) | 41.1 (40.1–42.0) | 36.9 (33.2–40.6) | |
| Unknown | 0.3 (0.2–0.4) | 0.3 (0.1–0.4) | 0.9 (0.1–1.6) | |
| Smoking status | Never | 49.9 (48.8–50.9) | 50.6 (49.5–51.7) | 39.1 (35.4–42.7) |
| Former | 38.9 (37.9–39.8) | 39.0 (37.9–40.0) | 37.3 (33.6–41.1) | |
| Current | 10.4 (9.8–11.0) | 9.9 (9.3–10.5) | 18.1 (15.3–21.0) | |
| Unknown | 0.8 (0.7–1.0) | 0.5 (0.4–0.7) | 5.5 (3.6–7.4) | |
| Overnight hospitalizations in past year | 0 | 82.4 (81.7–83.2) | 82.2 (81.5–83.0) | 85.2 (82.3–88.1) |
| 1 | 12.0 (11.4–12.7) | 12.1 (11.4–12.8) | 10.5 (8.1–13.0) | |
| 2+ | 5.6 (5.1–6.0) | 5.7 (5.2–6.1) | 4.3 (2.6–6.0) | |
The 11-item Schonberg mortality index takes into account multiple health factors including age, sex, body mass index, smoking, history of diabetes, emphysema, or cancer, hospitalizations in the past year, mobility, self-rated health, and functional limitation. Scores range from 0 to 25.[15,16]
IADL: instrumental activity of daily living
Adults without primary care were more likely to die over 15 years from any cause than those with primary care, especially those aged 70 and older. (Tables 2 and 3). Even after adjusting for estimated life expectancy and other factors associated with mortality in a Cox proportional hazards model, use of primary care was associated with a lower 15-year mortality risk (aHR: 0.84, 95% CI: 0.72–0.98) (Table 3). Median survival time was at least 2.1 years longer among those who used primary care (>15 years) compared to those who did not (12.9 years), with most individuals with primary care living beyond 15 years (Table 4). An adjusted 15-year survival curve is shown in Figure 1.
Table 2:
15-year all-cause mortality by exposure to primary care among NHIS participants
| Primary care | |||
|---|---|---|---|
| % Death in 15 years (2000+2005 samples) (95% CI) | Overall n=10,873 | Yes n=10,134 | No n=739 |
| Overall | 31.3 (30.3–32.2) | 31.1 (30.1–32.0) | 34.3 (31.1–37.5) |
| 65–69 (n=3266) | 21.4 (19.9–22.8) | 21.4 (19.8–22.9) | 21.2 (17.0–25.4) |
| 70–74 (n=3052) | 29.4 (27.7–31.2) | 29.0 (27.2–30.8) | 35.5 (29.7–41.3) |
| 75–79 (n=2610) | 37.8 (35.8–39.8) | 37.5 (35.5–39.5) | 43.2 (35.7–50.6) |
| 80–84 (n=1945) | 41.0 (38.6–43.3) | 40.6 (38.2–43.0) | 48.1 (38.4–57.9) |
Table 4:
Median survival (years) by primary care access and age group among NHIS participants
| Base modela | Primary care | |
|---|---|---|
| Yes | No | |
| Overall | >15 | 12.9 |
| Base model + age | ||
| Overall | >15 | 13.1 |
| Age 65–74 | >15 | 14.2 |
| Age 75–84 | 14.7 | 11.7 |
The base model included exposure to primary care, race/ethnicity, education, family income, health insurance, region, and the Schonberg index.
Figure 1: 15-year survival by exposure to primary care adjusted for covariatesa.
a. Median survival time, defined as the time at which estimated survival probability was equal to 0.5, was at least 2.1 years longer among those who used primary care (>15 years) compared to those who did not (12.9 years), with most individuals with primary care living beyond 15 years.
In addition to primary care, Asian race (aHR: 0.60, 95% CI: 0.45–0.80), Black race (HR: 0.80, 95% CI: 0.70–0.92), and Hispanic ethnicity (aHR: 0.75, 95% CI: 0.63–0.89) were associated with lower risk of 15-year mortality. Having less than high school education (aHR: 1.14, 95% CI: 1.03–1.27) and lower incomes were associated with increased risk of mortality.
We found that receipt of preventive services (colorectal cancer screenings in adults aged 65–75, flu and pneumonia vaccinations) and healthy behaviors (moderate/vigorous exercise) were more common in patients who reported having primary care (Supplemental Table). In the mediation analysis, the association between primary care and mortality, controlling for estimated life expectancy and sociodemographic factors in the base model, remained significant after adjusting for the receipt of flu/pneumonia vaccinations and moderate/vigorous exercise (aHR 0.82, 95% CI: 0.70–0.95). However, this relationship was no longer significant when colon cancer screening was added to the model (aHR: 0.89, 95% CI 0.72–1.10) (Table 5). This is likely because our sample was smaller when considering colon cancer screening, which only included adults aged 65–75 in the analysis. The smaller sample may have reduced our power to show an impact of primary care on mortality.
Table 5:
Mediation analysis
| Exposure to primary care | Added variable | ||||
|---|---|---|---|---|---|
| N | HR (95% CI) | p | HR (95% CI) | p | |
| Base model: Primary care, race, education, insurance, life expectancy, region, family income | 10,245 | 0.84 (0.72–0.98) | 0.03 | - | - |
| Base model + colon cancer screening (age 65–75) | 5,960 | 0.92 (0.75–1.14) | 0.46 | 0.86 (0.73–1.01) | 0.07 |
| Base model + flu vaccine | 10,245 | 0.83 (0.71–0.96) | 0.02 | 1.13 (1.04–1.22) | 0.004 |
| Base model + pneumonia vaccine | 10,245 | 0.85 (0.73–0.99) | 0.04 | 0.96 (0.89–1.04) | 0.34 |
| Base model + flu vaccine + pneumonia vaccine | 10,245 | 0.83 (0.71–0.97) | 0.02 | F: 1.17 (1.07–1.28) P: 0.91 (0.83–0.99) |
F: <0.001 P: 0.02 |
| Base model + vigorous exercise | 10,159 | 0.84 (0.72–0.98) | 0.03 | 0.97 (0.87–1.08) | 0.61 |
| Base model + moderate exercise | 10,058 | 0.83 (0.71–0.96) | 0.02 | 0.94 (0.87–1.02) | 0.15 |
| Base model + colon cancer screening, flu vaccine, pneumonia vaccine, vigorous exercise, moderate exercise | 5,832 | 0.89 (0.72–1.10) | 0.28 | - | - |
| Base model + flu vaccine, pneumonia vaccine, vigorous exercise, moderate exercise | 10,026 | 0.82 (0.70–0.95) | 0.01 | - | - |
DISCUSSION
This retrospective cohort study aimed to examine the association between having a usual provider, who offers services consistent with primary care, and mortality among older adults. US adults aged 65 to 84 who had a usual source of primary care had a lower risk of 15-year all-cause mortality than those who did not, with a 16% hazard reduction. Median life expectancy was also at least 2.1 years longer for individuals with primary care after adjusting for sociodemographic factors.
Our results add to previous research that has found increased PCP supply to be associated with decreased mortality and increased life expectancy at the population level.[13–15] We quantify this mortality benefit for individuals, demonstrating that older adults can expect to live approximately 766 days longer on average with primary care use. Our finding that primary care use increases individual rates of cancer screening, immunizations, and healthy behaviors suggests specific mechanisms through which primary care may impact life expectancy. However, in a mediation analysis, the association between primary care and mortality remained significant even after adjusting for several of these factors, such as vaccinations and exercise, suggesting that there may be other pathways that were not accounted for in our model, such as better control of cardiac risk factors (e.g., hypertension, hyperlipidemia, and diabetes) and earlier detection of cancer.[17,36,37] Additionally, primary care including geriatrics may facilitate longitudinal continuity of care, which has been associated with decreased all-cause mortality due to improved patient engagement, adherence to treatment, and timely recognition of clinical changes.[38–40] Our finding that diabetes and cancer were more common among participants with primary care may be reflecting that primary care leads to early detection and management of these conditions.
Taken together, although we cannot be certain of causality, our findings suggest that investments in primary care access could significantly improve the health and longevity of older adults. These investments are urgently needed. At present, an additional 13,075 PCPs are needed to satisfy minimum adequate population-to-PCP ratios, though the shortage is expected to grow to 87,150 PCPs by 2037.[2] Moreover, there are well-documented shortages of both geriatrician physicians and geriatrician nurse practitioners, with one study finding that 63.9% of US counties are without any provider of either type.[8] Several recommendations have been put forward to ensure that timely, high quality primary care is available to all.
Of note, the National Academies of Sciences, Engineering, and Medicine (NASEM) released a landmark report in 2021 which focused on improving access to high quality primary care.[41] Some recommendations included increasing funding for primary care through Medicare and Medicaid, reforming payment policy to invest more in primary care through fee-for-service and value-based payments, generating new care delivery models, decreasing administrative burdens, and investing in novel training approaches that encourage students to pursue primary care.[41] The Department of Health and Human Services has since launched the Initiative to Strengthen Primary Care and the Secretary’s Council on Primary Care to facilitate the implementation of these guidelines and best practices.[42] Several states have also passed legislation to boost investments in primary care, although additional work will be needed to meet current and projected shortages.[43]
In our analysis, Black race was associated with lower 15-year mortality risk in the adjusted Cox proportional hazards model. This finding aligns with recent studies suggesting that the increased mortality in Black individuals may diminish or even reverse later in life, often referred to as the “Black-White Mortality Cross-Over”.[44] This may also be due to selective mortality or the fact that our model adjusted for many of the mediators that may explain the increased mortality observed in Black individuals including health (e.g., smoking, obesity, history of diabetes or other diseases) and sociodemographic (e.g., education, family income, health insurance) factors. Similarly, Hispanic origin was associated with lower mortality risk, consistent with the “Hispanic Mortality Paradox.” This may be due to a selective migration advantage for healthier immigrants, differences in health behaviors, or sociocultural factors.[45,46] Finally, our finding that Asian populations had the lowest mortality risk aligns with previous studies documenting longer life expectancies.[25]
This study has a number of limitations. First, only 6.4% of individuals aged 65 to 84 reported not having a usual source of care in our sample; however, the large number of patients included over multiple years provided enough data to achieve statistical significance with relatively narrow confidence intervals. Second, our definition of primary care was based on self-report at a single point in time and relied on available survey data. As such, it was subject to inaccurate self-reporting and misclassification. It was also not possible to discern the amount of primary care received (e.g., frequency of visits, level of patient engagement); the quality of care provided (e.g., with respect to the core primary care functions of comprehensiveness and coordination); the type of PCP delivering care (e.g., medical specialty, physician vs. advanced practice provider); or whether patients maintained their level of access over the 15-year follow-up using available NHIS data. Third, because patient characteristics were assessed only at baseline and some factors (e.g., health insurance) may change over time and influence access to care, there was potential for residual confounding. Finally, while our study controlled for several sociodemographic and health-related factors, using a validated mortality index that included past medical history for several chronic conditions, certain variables associated with mortality such as total disease burden, disease severity, and rurality were not available in NHIS suggesting that unmeasured confounders could have affected our results.
Future research should examine whether differing levels of patient engagement in primary care and the resources available within a primary care practice lead to different mortality benefits. For example, studies could examine how different levels of patient engagement, different team-based approaches, and different types of services, such as care management, behavioral health integration, and population health, influence these outcomes. Future research could also examine the effect of primary care on older patients’ “health span” rather than life span, focusing on the quality of life gained to ensure that interventions extend life rather than prolong suffering.[47] While this study was limited by available data, other data sources, such as Medicare claims data, may hold promise in exploring these topics. Finally, questions remain about the generalizability of study findings. The US primary care system has several distinct characteristics, including a relative lack of formal gatekeeping structures and growing fragmentation, which may not apply to other countries.[21] Future research could examine the effect of primary care on mortality in different health settings.
CONCLUSIONS
Use of primary care was associated with decreased 15-year all-cause mortality risk among adults aged 65 to 84 in the United States. Alleviating PCP shortages and ensuring adequate access to primary care may be important for improving health outcomes among the aging population.
Supplementary Material
Funding
Dr. Schonberg’s effort was supported by a NIH/NIA K24 AG071906. The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.
Footnotes
Disclosures and Conflicts of Interest
Dr. Phillips is an advisor to Grow Therapy and Bicycle Health. He has grant support from AHRQ, the Commonwealth Fund, and the California Health Care Foundation, all of which are focused on primary care payment reform, but not on the topic of this manuscript.
Human Ethics and Consent to Participate
Not applicable.
Data Sharing
The data in this study can be obtained from the US National Center for Health Statistics: https://www.cdc.gov/nchs/nhis/index.html
References
- 1.Lawson E The Global Primary Care Crisis. Br J Gen Pract. 2022;73(726):3. Published 2022 Dec 21. doi: 10.3399/bjgp23X731469 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.National Center for Health Workforce Analysis. State of Primary Care Workforce, 2024. Nov 2024. https://bhw.hrsa.gov/sites/default/files/bureau-health-workforce/state-of-the-primary-care-workforce-report-2024.pdf [Google Scholar]
- 3.National Association of Community Health Centers. Closing the Primary Care Gap: How Community Health Centers Can Address the Nation’s Primary Care Crisis. Feb 2023. https://www.nachc.org/wp-content/uploads/2023/06/Closing-the-Primary-Care-Gap_Full-Report_2023_digital-final.pdf [Google Scholar]
- 4.Administration for Community Living. 2021 Profile of Older Americans. Nov 2022. https://acl.gov/sites/default/files/Profile%20of%20OA/2021%20Profile%20of%20OA/2021ProfileOlderAmericans_508.pdf [Google Scholar]
- 5.Huffstetler A, Topmiller M. Primary Care Access in Areas of the United States With Aging Populations. Am Fam Physician. 2024;109(4):308–309. [PubMed] [Google Scholar]
- 6.Moneme AN, Wirtalla CJ, Roberts SE, Keele LJ, Kelz RR. Primary Care Physician Follow-Up and 30-Day Readmission After Emergency General Surgery Admissions. JAMA Surg. 2023;158(12):1293–1301. doi: 10.1001/jamasurg.2023.4534 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gunta SP, Ul-Ejaz A, Murphy AM, et al. Association of number of primary care physicians with preventable hospitalizations and premature deaths. Postgrad Med. 2022;134(2):205–209. doi: 10.1080/00325481.2021.2021038 [DOI] [PubMed] [Google Scholar]
- 8.Xue Y, Poghosyan L, Lin Q. Supply and Geographic Distribution of Geriatric Physicians and Geriatric Nurse Practitioners. JAMA Netw Open. 2024;7(11):e2444659. Published 2024 Nov 4. doi: 10.1001/jamanetworkopen.2024.44659 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Shi L, Starfield B, Politzer R, Regan J. Primary care, self-rated health, and reductions in social disparities in health. Health Serv Res. 2002;37(3):529–550. doi: 10.1111/1475-6773.t01-1-00036 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Starfield B, Shi L, Macinko J. Contribution of primary care to health systems and health. Milbank Q 2005;83(3):457–502. doi: 10.1111/j.1468-0009.2005.00409.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Macinko J, Starfield B, Shi L. Quantifying the health benefits of primary care physician supply in the United States. Int J Health Serv. 2007;37(1):111–126. doi: 10.2190/3431-G6T7-37M8-P224 [DOI] [PubMed] [Google Scholar]
- 12.Friedberg MW, Hussey PS, Schneider EC. Primary care: a critical review of the evidence on quality and costs of health care. Health Aff (Millwood). 2010;29(5):766–772. doi: 10.1377/hlthaff.2010.0025 [DOI] [PubMed] [Google Scholar]
- 13.Basu S, Phillips RS, Berkowitz SA, Landon BE, Bitton A, Phillips RL. Estimated Effect on Life Expectancy of Alleviating Primary Care Shortages in the United States. Ann Intern Med. 2021;174(7):920–926. doi: 10.7326/M20-7381 [DOI] [PubMed] [Google Scholar]
- 14.Sharma A, Basu S. Does Primary Care Availability Mediate the Relationship Between Rurality and Lower Life Expectancy in the United States?. J Prim Care Community Health. 2022;13:21501319221125471. doi: 10.1177/21501319221125471 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Basu S, Berkowitz SA, Phillips RL, Bitton A, Landon BE, Phillips RS. Association of Primary Care Physician Supply With Population Mortality in the United States, 2005–2015. JAMA Intern Med. 2019;179(4):506–514. doi: 10.1001/jamainternmed.2018.7624 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chang CH, O’Malley AJ, Goodman DC. Association between Temporal Changes in Primary Care Workforce and Patient Outcomes. Health Serv Res. 2017;52(2):634–655. doi: 10.1111/1475-6773.12513 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sterling MR, Ferranti EP, Green BB, et al. The Role of Primary Care in Achieving Life’s Essential 8: A Scientific Statement From the American Heart Association. Circ Cardiovasc Qual Outcomes. 2024;17(12):e000134. doi: 10.1161/HCQ.0000000000000134 [DOI] [PubMed] [Google Scholar]
- 18.Rust G, Ye J, Baltrus P, Daniels E, Adesunloye B, Fryer GE. Practical barriers to timely primary care access: impact on adult use of emergency department services. Arch Intern Med. 2008;168(15):1705–1710. doi: 10.1001/archinte.168.15.1705 [DOI] [PubMed] [Google Scholar]
- 19.Grossman E, Legedza AT, Wee CC. Primary care for low-income populations: comparing health care delivery systems. J Health Care Poor Underserved. 2008;19(3):743–757. doi: 10.1353/hpu.0.0050 [DOI] [PubMed] [Google Scholar]
- 20.Gumas ED, Lewis C, Horstman C, Gunja MZ. Finger on the Pulse: The State of Primary Care in the U.S. and Nine Other Countries. Mar 2024. https://www.commonwealthfund.org/publications/issue-briefs/2024/mar/finger-on-pulse-primary-care-us-nine-countries [Google Scholar]
- 21.Penm J, MacKinnon NJ, Strakowski SM, Ying J, Doty MM. Minding the Gap: Factors Associated With Primary Care Coordination of Adults in 11 Countries. Ann Fam Med. 2017;15(2):113–119. doi: 10.1370/afm.2028 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Schonberg MA, Davis RB, McCarthy EP, Marcantonio ER. Index to predict 5-year mortality of community-dwelling adults aged 65 and older using data from the National Health Interview Survey. J Gen Intern Med. 2009;24(10):1115–1122. doi: 10.1007/s11606-009-1073-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Schonberg MA, Davis RB, McCarthy EP, Marcantonio ER. External validation of an index to predict up to 9-year mortality of community-dwelling adults aged 65 and older. J Am Geriatr Soc. 2011;59(8):1444–1451. doi: 10.1111/j.1532-5415.2011.03523.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Schonberg MA, Li V, Marcantonio ER, Davis RB, McCarthy EP. Predicting Mortality up to 14 Years Among Community-Dwelling Adults Aged 65 and Older. J Am Geriatr Soc. 2017;65(6):1310–1315. doi: 10.1111/jgs.14805 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.GBD US Health Disparities Collaborators. Life expectancy by county, race, and ethnicity in the USA, 2000–19: a systematic analysis of health disparities. Lancet. 2022;400(10345):25–38. doi: 10.1016/S0140-6736(22)00876-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Sasson I, Hayward MD. Association Between Educational Attainment and Causes of Death Among White and Black US Adults, 2010–2017. JAMA. 2019;322(8):756–763. doi: 10.1001/jama.2019.11330 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Chetty R, Stepner M, Abraham S, et al. The Association Between Income and Life Expectancy in the United States, 2001–2014 [published correction appears in JAMA. 2017 Jan 3;317(1):90. doi: 10.1001/jama.2016.18691.]. JAMA. 2016;315(16):1750–1766. doi: 10.1001/jama.2016.4226 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Woolhandler S, Himmelstein DU. The Relationship of Health Insurance and Mortality: Is Lack of Insurance Deadly?. Ann Intern Med. 2017;167(6):424–431. doi: 10.7326/M17-1403 [DOI] [PubMed] [Google Scholar]
- 29.Dwyer-Lindgren L, Bertozzi-Villa A, Stubbs RW, et al. Inequalities in Life Expectancy Among US Counties, 1980 to 2014: Temporal Trends and Key Drivers. JAMA Intern Med. 2017;177(7):1003–1011. doi: 10.1001/jamainternmed.2017.0918 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Centers for Disease Control and Prevention. 2000 Imputed Family Income/Personal Earnings Files. Jul 2021. https://archive.cdc.gov/www_cdc_gov/nchs/nhis/2000imputedincome.htm [Google Scholar]
- 31.Centers for Disease Control and Prevention. 2000 Supplemental Imputed Family Income/Personal Earnings Files. Nov 2015. https://archive.cdc.gov/www_cdc_gov/nchs/nhis/2000supplementalimputedincome.htm [Google Scholar]
- 32.Buchman TG, Simpson SQ, Sciarretta KL, et al. Seasonal influenza vaccination is associated with reduced risk of death among Medicare beneficiaries☆. Vaccine. 2021;39(52):7569–7577. doi: 10.1016/j.vaccine.2021.11.016 [DOI] [PubMed] [Google Scholar]
- 33.Chung H, Buchan SA, Campigotto A, et al. Influenza Vaccine Effectiveness Against All-Cause Mortality Following Laboratory-Confirmed Influenza in Older Adults, 2010–2011 to 2015–2016 Seasons in Ontario, Canada. Clin Infect Dis. 2021;73(5):e1191–e1199. doi: 10.1093/cid/ciaa1862 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Martinez-Gomez D, Luo M, Huang Y, et al. Physical Activity and All-Cause Mortality by Age in 4 Multinational Megacohorts [published correction appears in JAMA Netw Open. 2025 Jan 2;8(1):e2460155. doi: 10.1001/jamanetworkopen.2024.60155.]. JAMA Netw Open. 2024;7(11):e2446802. Published 2024 Nov 4. doi: 10.1001/jamanetworkopen.2024.46802 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Hupin D, Roche F, Gremeaux V, et al. Even a low-dose of moderate-to-vigorous physical activity reduces mortality by 22% in adults aged ≥60 years: a systematic review and meta-analysis. Br J Sports Med. 2015;49(19):1262–1267. doi: 10.1136/bjsports-2014-094306 [DOI] [PubMed] [Google Scholar]
- 36.Laddu D, Ma J, Kaar J, et al. Health Behavior Change Programs in Primary Care and Community Practices for Cardiovascular Disease Prevention and Risk Factor Management Among Midlife and Older Adults: A Scientific Statement From the American Heart Association. Circulation. 2021;144(24):e533–e549. doi: 10.1161/CIR.0000000000001026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Qiao EM, Guram K, Kotha NV, et al. Association Between Primary Care Use Prior to Cancer Diagnosis and Subsequent Cancer Mortality in the Veterans Affairs Health System. JAMA Netw Open. 2022;5(11):e2242048. Published 2022 Nov 1. doi: 10.1001/jamanetworkopen.2022.42048 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Leleu H, Minvielle E. Relationship between longitudinal continuity of primary care and likelihood of death: analysis of national insurance data. PLoS One. 2013;8(8):e71669. Published 2013 Aug 22. doi: 10.1371/journal.pone.0071669 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Baker R, Freeman GK, Haggerty JL, Bankart MJ, Nockels KH. Primary medical care continuity and patient mortality: a systematic review. Br J Gen Pract. 2020;70(698):e600–e611. Published 2020 Aug 27. doi: 10.3399/bjgp20X712289 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Yen PC, Lo YT, Lai CC, et al. Effectiveness of outpatient geriatric evaluation and management intervention on survival and nursing home admission: a systematic review and meta-analysis of randomized controlled trials. BMC Geriatr. 2023;23(1):414. Published 2023 Jul 7. doi: 10.1186/s12877-023-04036-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.National Academies of Sciences, Engineering, and Medicine. Implementing High-Quality Primary Care: Rebuilding the Foundation of Healthcare. 2021. https://nap.nationalacademies.org/catalog/25983/implementing-high-quality-primary-care-rebuilding-the-foundation-of-health [PubMed] [Google Scholar]
- 42.Hughes LS, Cohen DJ, Phillips RL Jr. Strengthening Primary Care to Improve Health Outcomes in the US-Creating Oversight to Address Invisibility. JAMA Health Forum. 2022;3(9):e222903. Published 2022 Sep 2. doi: 10.1001/jamahealthforum.2022.2903 [DOI] [PubMed] [Google Scholar]
- 43.Bodenheimer T Revitalizing Primary Care, Part 2: Hopes for the Future. Ann Fam Med. 2022;20(5):469–478. doi: 10.1370/afm.2859 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Breen C Black-White Mortality Crossover: New Evidence from Social Security Mortality Records. doi: 10.31235/osf.io/ax9u3. [DOI] [Google Scholar]
- 45.Ruiz JM, Steffen P, Smith TB. Hispanic mortality paradox: a systematic review and meta-analysis of the longitudinal literature. Am J Public Health. 2013;103(3):e52–e60. doi: 10.2105/AJPH.2012.301103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Nicole W Paradox Lost? The Waning Health Advantage among the U.S. Hispanic Population. Environ Health Perspect. 2023;131(1):12001. doi: 10.1289/EHP11618 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Crimmins EM. Lifespan and Healthspan: Past, Present, and Promise. Gerontologist. 2015;55(6):901–911. doi: 10.1093/geront/gnv130 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data in this study can be obtained from the US National Center for Health Statistics: https://www.cdc.gov/nchs/nhis/index.html