Principal FindingsWe have shown that a wide range of measures of health care utilization display significant and strong associations with the risk of first-time LTC need. To the best of our knowledge, our study is the first to analyze the interplay of the frequency of physician visits, DMP enrollment, physiotherapy utilization, prescription of medical aids, screenings and vaccinations, and inadequate medication supply with first-time LTC need simultaneously. In the population aged ≥60 years, utilization of specialist care, physiotherapy, and screenings and vaccinations display the strongest relationship with nonoccurrence of first-time LTC need. At the same time, frequent utilization of GP care, hospitalizations, polypharmacy, and prescription of walking aids showed a significant association with first-time LTC need. As this study relied on a 5-year exposure period to evaluate health care utilization, the relationship between health care use in a 5-year period and LTC need can only be assessed in and does only apply to individuals who neither need LTC nor die within 5 years. This approach provides a longer-term perspective on the health care trajectory of a person, which may be more meaningful for prevention than a single event.
Comparison With Prior WorkApart from the 2 studies, which corroborated that inadequate medication supply is related to an increase in future LTC risk [10,11], further evidence on the relationship of health care utilization with first-time LTC need is missing. LTC need being a sociolegal concept in Germany that differs substantially from that in other countries [3,26], prior research has focused on related outcomes, such as frailty, functional measures, morbidity, or mortality [8,27,28]. Although these outcomes must be distinguished both from each other and first-time LTC need, they are interrelated [29-32] and evidence on how they correlate with health care utilization can promote our understanding of the link between health care utilization and first-time LTC need.
Frequency of Physician ContactWe evaluated 3 variables that are closely related to the frequency of physician contact: GP billing days, specialist physician billing days, and hospital admissions. Compared to no contact, both GP billing days and hospital admissions were related to an increased risk of first-time LTC need. While utilization of specialist care with a low number of billing days and specialist groups showed a preventive association, all 3 variables have in common that compared to a low level of utilization, a higher level of physician contact was related to an increase in the risk of first-time LTC need. On the one hand, frequent physician contact may entail constant assessment of health risks and subsequent counseling and thereby exerting a preventive effect on LTC need risk [33,34]. On the other hand, there is a wealth of literature on how physical frailty is associated with a higher utilization of health care in terms of outpatient care, inpatient care, home care, and nursing home admission [35-40]. Moreover, individuals with an adverse status in terms of living alone, social relations, and social support, referred to as socially frail, are reported to have a lower number of physician visits and lower likelihood of hospitalization, hinting at lower ability to access health care services including LTC [41]. Despite our efforts to control for comorbidities, both physical and social frailty likely act as confounders, correlating both with first-time LTC need risk and health care utilization.
In sensitivity analyses that excluded the last year from the exposure period or extended follow-up to 1 year, we also observed a markedly weaker association between hospitalizations and LTC need. This is consistent with the fact that hospitalizations frequently occur shortly before first-time LTC need [42]. If there is a time lag between the end of the exposure period and the date of first-time LTC need, hospitalizations shortly before first-time LTC need will not be reflected in the exposure variable. This underscores how the strong relationship between hospitalizations and LTC need in the main analysis reflects morbidity levels. In summary, while the frequency of physician contact is a valuable identifier of persons at risk of first-time LTC need, it cannot be assumed to be detrimental to independent living based on our analysis.
Condition-Specific Health Care UtilizationApart from the frequency of physician contact, we also evaluated how extensive specialist outpatient care utilization was by including the number of specialist groups consulted. At a constant number of specialist billing days, an increase in specialist groups consulted was connected to a decreased risk of first-time LTC need. Similarly, another claims data–based study from Germany reported that, adjusted for GP care utilization, LTC beneficiaries are less likely to visit physicians from different specialist groups than older people without LTC need [43]. We cannot rule out that part of the promising association we estimated pertains to barriers to accessibility of specialist care. These might be more pronounced for individuals at higher risk of first-time LTC need. Still, hypothesizing that even with few patient-physician contacts, each specialist group can give specific preventive recommendations and guidance, it is plausible that a wider variety of specialist groups consulted, holding the number of billing days fixed, exerts a preventive effect. On the other hand, having visited a given specialist more frequently may reflect higher morbidity levels more so than an improved capacity to provide preventive guidance.
Of the 4 DMPs evaluated, we found a preventive relationship consistent across main and sensitivity analyses regarding only the congenital heart disease DMP, while the COPD DMP even showed a detrimental association. As we can only account for disease presence and not severity, selection effects might bias our estimates if DMP participants differ from nonparticipants in this regard. DMPs have not yet been evaluated in terms of their association with first-time LTC need, nor have there been DMP evaluations using randomized control trials in Germany [44].
While we found physiotherapy utilization to be predictive of nonoccurrence of first-time LTC need, when receiving prescriptions in up to 10 quarters within 5 years, we saw no such association for higher levels of utilization. There is prior evidence that physiotherapy can support the use of medical aids [45], prevent institutionalization after hospitalization [46], and, in the case of a Spanish study, reduce mortality through improvement in quality of life, alleviation of musculoskeletal disorders, a referral function in the health care system, and fall prevention [47]. While there is currently no evidence on the relationship between physiotherapy utilization and first-time LTC need, 2 articles report that frail persons receive physiotherapy more often than nonfrail ones [36,38].
Furthermore, our medical aids analysis showed that only orthopedic devices had a negative relationship with first-time LTC need. All other categories except for hearing aids had a positive association that was most pronounced in walking aids. A meta-analysis suggested that insoles, which make up for the largest share of prescriptions within the orthopedic devices group, were beneficial for balance improvement and fall prevention [48], supporting our finding of a preventive potential. Although walking aids are intended to prevent falls, walking aid users face a high risk of falls and injuries that might be reduced by improvements in devices and better instructions on device usage [45,49]. Likely, both difficulties in accounting for confounders such as gait quality and potential for improvement in walking aid use contribute to the strong positive association with first-time LTC need we observe.
Screenings and VaccinationsThe cancer screenings, general health check, and influenza vaccination summarized in a single variable displayed a strong relationship with nonoccurrence of first-time LTC need. There is tentative evidence supporting the health benefits of these services in the population aged ≥60 years, yet the association we find remains high against this backdrop.
Despite the potential of cancer screening to reduce all-cause mortality by 1% to 3% in Germany [50], screening cessation may be recommended for prostate cancer, cervical cancer, and colorectal cancer. With increasing age and decreasing life expectancy, potential harms in screening may outweigh potential benefits, necessitating an individualized screening decision [51-53]. On the one hand, this calls into question whether cancer screenings can have a substantial preventive effect in terms of LTC need. On the other hand, it raises the question of whether part of the observed effect reflects screening cessation recommendations being put into practice. General health checks and health risk assessments have been shown to increase uptake of other preventive services, but an effect on mortality cannot be substantiated despite positive impact on chronic diseases treatment and patient-reported outcomes [34,54-56]. The influenza vaccination provides the most concrete reference for a potential delaying effect on first-time LTC need. Not only is it crucial to influenza prevention, especially so in older adults [57], but it is also related to a decreased risk of dementia [58,59], the third most-named diagnosis for first-time LTC need in Germany in 2021 [60].
Other than effects of screenings and vaccinations themselves, preventive health behavior could contribute to the strength of the observed association. It has been shown that the uptake of preventive services is higher in individuals with health-promoting behavior [61,62]. As we have no information on crucial behavioral aspects such as physical activity, nutrition, and alcohol and tobacco use, they likely bias the observed relationship toward prevention. Bearing that in mind, our study provides first hints for a possible delaying effect of screenings and vaccinations on first-time LTC need that further studies should investigate in more detail on a broader base of data.
Inadequate Medication SupplyOur finding that inadequate medication supply is positively related to first-time LTC need and polypharmacy, more so than PIM, is in line with prior cross-sectional [63] and longitudinal studies [10,11] from Japan. Kuroda et al [10] evaluated polypharmacy and PIM using drug prescriptions from a 2-year exposure period and reported associations of similar magnitude. Inadequate medication supply has been shown to be associated with falls, dementia, functional and cognitive decline, and sarcopenia [28,64,65], all of which are plausible causes of first-time LTC need. This suggests that correction of inadequate medication supply can be a leverage point to LTC need prevention.
Strengths and LimitationsThis retrospective cohort study adheres to certain conditions necessary to justify causal statements about the observed associations. Outcome was evaluated after exposure, and thus, reverse causality through effects of utilization of LTC insurance services can be ruled out. Still, utilization might be driven by morbidity trends on a trajectory approaching first-time LTC need [42]. Excluding the final year of the exposure period from the analysis showed that only estimates related to hospitalization are affected strongly by such morbidity trends. The robustness of the remaining estimates indicates that reverse causality through short-term morbidity trends is no major concern. The large nationwide sample enabled us to assess utilization of multiple health care services simultaneously despite their interrelation. In this way, we were able to measure the relationship of each utilization variable with first-time LTC need given the utilization pattern in terms of the other variables.
Nonetheless, this study faces some substantial limitations that confine the results to the domain of correlation. It is the nature of claims data that they lack information on disease severity, health-promoting behavior, social networks, and individual-level socioeconomic variables. These parameters need to be integrated to enable attainment of unbiased estimates, given that they are related to physical and social frailty, as well as the accessibility of health care [66,67]. Furthermore, despite the large sample size, results attained based on AOK-insured individuals cannot be generalized to the German population, as they tend to have lower socioeconomic status and higher comorbidity burden than people insured privately or with other SHI funds [68,69]. On top of that, our observation period coincides with the COVID-19 pandemic, whose effect on health care utilization limits the external temporal validity of our study.
Moreover, a range of health care services that may contribute to LTC need prevention, such as breast cancer screening, geriatric assessment, and rehabilitation, were not evaluated as part of this study. This study was designed to estimate associations of health care utilization with first-time LTC need without placing further restrictions on the study population than the criterion of being aged ≥60 years. Therefore, services such as breast cancer screening and geriatric assessment, access to which is specific to sex and age, could not be included. In Germany, rehabilitation services are covered by the statutory pension insurance scheme before retirement. Utilization of rehabilitation services can thus not be assessed on SHI claims data alone for a population partially still in employment. Furthermore, we primarily aimed to assess the association of health care utilization with first-time LTC need as a proxy for functional health. We must concede that claiming LTC benefits is itself an act of health care utilization that, besides functional health, is determined by access to health care and utilization behavior. As a determinant of health care accessibility and utilization behavior, individual-level socioeconomic variables could be a valuable addition in factoring out its confounding impact.