Skip to main content
Health Services Research logoLink to Health Services Research
. 2017 Aug 31;53(4):2227–2248. doi: 10.1111/1475-6773.12763

Multilevel Comparisons of Hospital Discharge among Older Adults with a Fall‐Related Hospitalization

Samuel D Towne Jr 1,, Kayla Fair 2, Matthew Lee Smith 1,3, Diane M Dowdy 1, SangNam Ahn 1,4, Obioma Nwaiwu 5, Marcia G Ory 6
PMCID: PMC6051977  PMID: 28857156

Abstract

Objective

We examined multilevel factors associated with hospital discharge status among older adults suffering a fall‐related hospitalization.

Data Sources

The 2011–2013 (n = 131,978) Texas Inpatient Hospital Discharge Public‐Use File was used.

Study Design/Methods

Multilevel logistic regression analyses estimated the likelihood of being discharged to institutional settings versus home.

Principal Findings

Factors associated with a greater likelihood of being discharged to institutional settings versus home/self‐care included being female, white, older, having greater risk of mortality, receiving care in a non‐teaching hospital, having Medicare (versus Private) coverage, and being admitted from a non‐health care facility (versus clinical referral).

Conclusions

Understanding risk factors for costly discharges to institutional settings enables targeted fall‐prevention interventions with identification of at‐risk groups and allows for identifying policy‐related factors associated with discharge status.

Keywords: Falls, hospital discharge, older adults, Medicare


Falls disproportionately affect older adults and when falls do occur, they result in serious injury at a much greater rate than among younger individuals (Sterling, O'Connor, and Bonadies 2001), thereby disproportionately affecting the already hard‐hit U.S. health care delivery and finance system. Those with a previous history of a fall are more likely to enter costly institutionalized settings, namely nursing homes (Tinetti and Williams 1997). Identifying factors associated with discharge location can help key stakeholders be better informed to identify the best solutions to prevent falls or a recurrent fall(s).

Some have suggested that focusing on post‐acute discharge may hold major implications for reducing overall spending (Das et al. 2016; Mor, Rahman, and McHugh 2016). Focusing on discharging patients to less costly locations (e.g., home) has been suggested as a major factor in this process of lowering hospital spending (Mor, Rahman, and McHugh 2016). Large variation in the price of medical care may be attributed to variation among hospitals based on various measures such as performance (Das et al. 2016). Thus, accounting for hospital‐level variation (e.g., using multilevel analyses) is critical (Rice and Leyland 1996). Ongoing surveillance of fall‐related hospitalization is needed to ensure the most current evidence is available to inform policy.

Aims

Although multiple studies have examined injuries related to serious falls (Masud and Morris 2001; Smith et al. 2010), few studies have investigated factors that may contribute to the discharge location of a patient following a traumatic fall (Lim, Hoffmann, and Brasel 2007). Therefore, our objective was to examine multilevel factors associated with hospital discharge status among older adults suffering a fall‐related hospitalization. We aimed to examine fall‐related hospitalization by (1) overall distribution, (2) discharge location, and (3) to identify multilevel factors associated with discharge to institutional settings versus to home or self‐care (routine discharge)—henceforth HSC or home health care (HHC). Multiple definitions of discharge to home were included given discharge to home with self‐care may incur different costs than being discharged to home with the inclusion of HHC.

Design and Methods

Data

The Texas Hospital Inpatient Discharge Public‐Use Data for 2011 (base‐file n = 2,937,634), 2012 (base‐file n = 2,965,961), and 2013 (base‐file n = 2,910,853) were used (THHS 2011).

Patient Population

The target population included older adults (age 65+), suffering a fall‐related hospitalization, and being admitted from a non–health care facility (NHCF; n = 40,342, 42,864, 44,977; 2011–2013, respectively) or a clinical referral (n = 3,200, 3,004, 3,221; 2011–2013, respectively). Excluding transfers from another facility (e.g., transfer from a hospital or other health care facility) allowed us to identify those most likely to come from a residential setting (e.g., those aging‐in‐place) versus institutional settings. These excluded observations represented 11.8 percent of fall‐related hospitalizations. Discharges with a primary payment source of “other non‐federal programs” (<0.5 percent for each year) were excluded due to the inability to identify specific sources of payment. The analytical group included 41,933, 43,989, and 46,056 in 2011, 2012, and 2013, respectively.

Main Outcome Measures

Discharge location was the dependent variable. Being discharged home without any additional care (self‐care) versus being discharged home with additional care (e.g., HHC) is associated with different medical costs. Thus, we separated discharge to one's home into three categories in separate analyses. Category A: All discharges to HSC, that is, home (i.e., self‐care/routine discharge), including formal service (i.e., HHC or intravenous therapy [IV]); Category B: Discharges to home without formal service (i.e., HHC or IV); Category C: Discharges to home with formal service (i.e., HHC or IV). Category A = B + C: All other discharges included discharges/transfers to other short‐term general hospital; skilled nursing facility; intermediate care facility; cancer center; admitted as inpatient to this hospital; still patient; federal health care facility; hospice–medical facility; within this institution to Medicare‐approved swing bed; inpatient rehabilitation facility; Medicare‐certified long‐term care hospital; Medicaid‐certified nursing facility; psychiatric hospital or psychiatric distinct part of a hospital; critical access hospital; other outpatient service; and institution outpatient.

Individuals were coded as expired/deceased; discharged to hospice–home; and left against medical advice, where excluded from analyses. Our adjusted model assessed fall‐related hospitalizations among those either coming from a non‐institutional setting or by way of clinical referrals (CR).

Covariates

Individual‐level factors included in the adjusted model were as follows: sex (male/female); race/ethnicity (non‐Hispanic American Indian/Alaska Native, non‐Hispanic Asian or Pacific Islander, non‐Hispanic black, non‐Hispanic white, and Hispanic); age group (65–74, 75–84, and 85+); payment source identified as the expected primary source of payment (Title‐V or other federal program; Veteran Administration plan or Civilian Health and Medical Program of the Uniformed Services, Medicare; Medicaid; charity, indigent or unknown; workers compensation; other forms of payment (non‐federal/non‐charity/non‐workers compensation) operationally defined as “private” in the current analyses (liability medical, liability, health maintenance organization, disability insurance, commercial insurance, Blue Cross Blue Shield, automobile medical, indemnity insurance, exclusive provider organization, point of service, preferred provider organization, and central certification); and risk of mortality (minor, moderate, major, or extreme risk upon admission).

Hospital‐ and community‐level factors were also included to account for differences at the hospital‐level and individual residence. Hospital‐level teaching status was included to account for differences in being a teaching‐affiliated facility versus a non‐teaching facility. Neighborhood‐level characteristics including classification as a large central metropolitan, large fringe metropolitan, medium metropolitan, small metropolitan, micropolitan, and noncore area were included to account for differences associated with rural versus urban residence using the National Center for Health Statistics Urban‐Rural Classification Scheme. Source of admission, coded as clinical referral (e.g., the patient was referred by a provider from an outpatient clinic including a physician at the hospital) versus NHCF (e.g., patient residing at home prior to admission), was also included to assess potential differences in patient discharge location/status. To further describe our patient population (descriptive statistics only), we identified ICD‐9 codes for the primary diagnosis upon admission and average length of stay (LOS).

Statistical Analyses

Analyses were conducted using SAS 9.4 using random coefficient models (RCMs) to assess the likelihood of our outcomes. In adjusted analyses, the payment source associated with the hospital discharge was collapsed into Medicare accounting for over 90 percent of all discharges, Private which accounted for between 5 and 7 percent of all cases, and Other given the relatively small analytical group size among other payers. Intraclass correlation coefficients (ICCs) were calculated using a model without predictors (Bell, Ene, and Schoeneberger 2013). The decision to use RCMs was made based on two factors: (1) the nested nature of the data where individuals are nested within hospitals fits with the theoretical framework of multilevel modeling; (2) the ICCs, while all <10 percent, did range approximately 4–9 percent, indicating some variation in our outcomes was likely attributable to differences among hospitals.

Ethical Approval of Studies and Informed Consent

Ethical approval was granted through the Texas A&M University Institutional Review Board (IRB).

Results

Table 1 presents the distribution of older adults admitted to the hospital for a fall‐related injury through non‐institutionalized settings (NIS) or CR (referred to collectively henceforth as NIS/CR) prior to admission by selected characteristics. Overall, the number of older adults suffering fall‐related hospitalizations who were admitted through NIS/CR prior to admission increased from 41,933 in 2011, to 43,989 in 2012, and then to 46,056 in 2013. When testing for significant differences in our binary outcomes of discharge location (i.e., 1 model for each category A, B, C), we find no evidence to suggest any significant difference (α = 0.01) by year. To further describe our patient population, we identified ICD‐9 codes for the primary diagnosis. When combining all 3 years of data, we find approximately a third (36 percent) of fall‐related hospitalizations were associated with a fracture, while most others were associated with infections (approximately 40 percent) and other diagnosis (e.g., circulatory or respiratory issues). The average LOS was approximately 5 days (range 1–368). When stratifying LOS by discharge location, we find the average LOS was 5.8 days for those discharged to an institutional setting versus 4.1 days (Category A), 3.7 days (Category B), and 4.9 days (Category C) for being discharged home, respectively.

Table 1.

Distribution of the Population by Selected Characteristics

2011 Fall‐Related Hospitalizations among Those Aged 65 and Older Being Admitted from Non‐Institutionalized Settings 2012 Fall‐Related Hospitalizations among Those Aged 65 and Older Being Admitted from Non‐Institutionalized Settings 2013 Fall‐Related Hospitalizations among Those Aged 65 and Older Being Admitted from Non‐Institutionalized Settings
N Percent N Percent N Percent
Discharge status Institution 27682 66.01 28879 65.65 30787 66.85
Category A Home (home and/or home health) 14251 33.99 15110 34.35 15269 33.15
Category B Home (routine care only) 9254 25.05 9646 25.04 9765 24.08
Category C Home health only 4997 15.29 5464 15.91 5504 15.17
Rurality Large central metropolitan 18332 43.00 19185 42.91 20044 42.68
Large fringe metropolitan 7237 16.98 7851 17.56 8518 18.14
Medium metropolitan 7455 17.49 8264 18.48 8499 18.10
Small metropolitan 3616 8.48 3373 7.54 3571 7.60
Micropolitan 3190 7.48 3225 7.21 3357 7.15
Noncore 2799 6.57 2810 6.29 2969 6.32
Sex Female 30236 68.97 31528 68.42 32830 67.76
Male 13604 31.03 14551 31.58 15619 32.24
Race/ethnicity American Indian/Alaska Native 411 0.98 348 0.83 108 0.23
Asian or Pacific Islander 439 1.04 433 1.03 748 1.62
Black 2060 4.89 2020 4.81 2309 5.01
White 31873 75.69 31240 74.36 34455 74.75
Hispanic 7329 17.40 7971 18.97 8474 18.38
Age group 65–74 years 10623 24.23 11463 24.88 12473 25.74
75–84 years 17189 39.21 17813 38.66 18674 38.54
85+ years 16029 36.56 16804 36.47 17303 35.71
Risk of mortality Minor 9705 22.14 9448 20.50 10779 22.25
Moderate 20108 45.87 20517 44.52 21256 43.88
Major 9960 22.72 11319 24.56 13007 26.85
Extreme 4068 9.28 4796 10.41 3404 7.03
Teaching status Non‐teaching 35152 80.21 36857 80.01 38418 79.34
Teaching 8673 19.79 9209 19.99 10003 20.66
Payment source TV Title‐V or other federal program 34 0.08 63 0.14 64 0.13
Veteran Administration Plan or CHAMPUS 147 0.34 147 0.32 189 0.39
Medicare 40395 92.86 42623 92.63 44068 91.06
Medicaid 278 0.64 238 0.52 273 0.56
Charity, indigent or unknown 96 0.22 571 1.24 485 1.00
Workers compensation health claim 152 0.35 130 0.28 145 0.30
Private 2399 5.51 2243 4.87 3168 6.55
Source of admission Clinical referral 3200 7.35 3004 6.55 3221 6.68
Non‐health care facility 40342 92.65 42864 93.45 44977 93.32
Transfer from health care facility Excluded from analyses Excluded from analyses Excluded from analyses Excluded from analyses Excluded from analyses Excluded from analyses

The proportion of older adults admitted for fall‐related injury through NIS/CR that were discharged to HSC or HHC was approximately 34 percent in 2011 and 2012, and 33 percent in 2013. The bulk of older adults suffering a fall‐related hospitalization admitted through NIS/CR prior to admission were in metropolitan areas. In addition, females accounted for the greatest proportion of fall‐related hospitalization, with 69 percent in 2011 and approximately 68 percent in 2012 and 2013. While the majority of older adults suffering a fall‐related hospitalization who were admitted through NIS/CR prior to admission were identified as white, nearly one in five were Hispanic individuals. Individuals aged 65–74 represented nearly one in four discharges among older adults suffering a fall‐related hospitalization who were admitted through NIS/CR.

Nearly half of older adults suffering a fall‐related hospitalization who were admitted through NIS/CR prior to admission were admitted with a moderate risk of mortality.

For 2011, 2012, and 2013, approximately 79–80 percent of older adults suffering a fall‐related hospitalization who were admitted through NIS/CR prior to admission were treated in non‐teaching institutions. Medicare represented over 91 percent of all payments sources for older adults suffering a fall‐related hospitalization who were admitted through NIS/CR. Ninety‐three percent of the patients in our study were admitted from the home with 7 percent being referred from the clinic.

Table 2 presents the distribution of older adults suffering a fall‐related hospitalization who were admitted through NIS/CR prior to admission. The largest segment of all discharges among older adults suffering a fall‐related hospitalization who were admitted through NIS/CR prior to admission was represented by discharge to skilled nursing care (approximately 35–37 percent).

Table 2.

Discharge Location/Status for Individuals Aged 65 and Older with a Fall‐Related Hospitalization in 2011, 2012, and 2013

2011 Fall‐Related Hospitalizations among Those Aged 65 and Older Being Admitted from Non‐Institutionalized Settings 2011 Total Fall‐Related Hospitalizations among Those Aged 65 and Older 2012 Fall‐Related Hospitalizations among Those Aged 65 and Older Being Admitted from Non‐Institutionalized Settings 2012 Total Fall‐Related Hospitalizations among Those Aged 65 and Older 2013 Fall‐Related Hospitalizations among Those Aged 65 and Older Being Admitted from Non‐Institutionalized Settings 2013 Total Fall‐Related Hospitalizations among Those Aged 65 and Older
N Percent N Percent N Percent N Percent N Percent N Percent
Discharged to home Home or self‐care (routine discharge) 9254 21.11 10133 20.55 9646 20.93 10524 20.28 9765 20.16 10729 19.64
Care of home health service 4993 11.39 5761 11.69 5455 11.84 6570 12.66 5499 11.35 6710 12.28
Care of Home IV provider 4 0.01 4 0.01 9 0.02 9 0.02 5 0.01 5 0.01
Discharged to other location (not home) Other short‐term general hospital 576 1.31 737 1.49 577 1.25 765 1.47 651 1.34 855 1.56
Skilled nursing facility 15364 35.04 17488 35.47 16472 35.75 18673 35.98 17743 36.62 20026 36.65
Intermediate care facility 924 2.11 1105 2.24 844 1.83 1032 1.99 941 1.94 1104 2.02
Designated cancer center 13 0.03 15 0.03 23 0.05 25 0.05 18 0.04 19 0.03
Admitted as inpatient to this hospital 3 0.01 3 0.01 7 0.02 10 0.02 10 0.02 11 0.02
Still patient 4 0.01 5 0.01 2 0.00 2 0.00 3 0.01 3 0.01
Discharged/transferred to federal health care facility 26 0.06 32 0.06 19 0.04 25 0.05 20 0.04 23 0.04
Hospice–medical facility 690 1.57 805 1.63 864 1.88 985 1.90 916 1.89 1050 1.92
Discharged/transferred within this institution to Medicare‐approved swing bed 305 0.70 385 0.78 399 0.87 459 0.88 335 0.69 397 0.73
Inpatient rehabilitation facility 8393 19.14 9044 18.35 8265 17.94 8858 17.07 8798 18.16 9449 17.29
Medicare‐certified long‐term care hospital 1130 2.58 1263 2.56 1141 2.48 1280 2.47 1096 2.26 1234 2.26
Medicaid‐certified nursing facility 192 0.44 254 0.52 196 0.43 242 0.47 179 0.37 212 0.39
Psychiatric hospital or psychiatric distinct part of a hospital 55 0.13 65 0.13 64 0.14 75 0.14 67 0.14 81 0.15
Critical Access Hospital (CAH) 7 0.02 9 0.02 6 0.01 6 0.01 10 0.02 10 0.02
Other outpatient service 0 0 0 0 0 0 0 0 0 0 0 0
Institution outpatient 0 0 0 0 0 0 0 0 0 0 0 0
Other (Excluded from adjusted analyses) Expired 1244 2.84 1453 2.95 1296 2.81 1495 2.88 1368 2.82 1588 2.91
Expired at home 0 0 0 0 0 0 0 0 0 0 0 0
Expired in a medical facility 1 0.00 1 0.00 8 0.02 8 0.02 7 0.01 8 0.01
Expired, place unknown 0 0 0 0 0 0 0 0 0 0 0 0
Hospice–home 485 1.11 549 1.11 583 1.27 626 1.21 708 1.46 775 1.42
Left against medical advice 93 0.21 98 0.20 112 0.24 122 0.24 129 0.27 142 0.26

Adjusted Analyses

Discharged to HSC or HHC

Table 3 presents results for the likelihood of being discharged to an institutional health care setting versus HSC or HHC. Factors associated with (α = 0.05) a greater likelihood of being discharged to institutionalized settings among older adults suffering a fall‐related hospitalization who were admitted through NIS/CR included the following: being female; being in older age groups versus those aged 65–74; having a higher risk of mortality versus minor risk; being treated in a non‐teaching facility versus a teaching facility; and having Medicare as the primary source of payment, after adjusting for all other terms in the model. In contrast, factors associated with (α = 0.05) a lower likelihood of being discharged to an institutionalized health care setting included the following: being an Asian or Pacific Islander, black, or Hispanic individual versus being a white individual; having the primary source of payment listed as Private versus Medicare; and being admitted from a clinical referral versus a NHCF after controlling for all other terms in the model across all years under study. In 2011, residents of the most metropolitan areas known as large central metropolitan areas were more likely to be discharged to institutionalized settings versus those residing in medium metropolitan or small metropolitan areas, after adjusting for all other terms in the model.

Table 3.

Adjusted Analyses Predicting Discharge to an Institutional Setting versus Home/Routine Care or Home Health Care (HHC)

2011 2012 2013
95% Confidence Interval 95% Confidence Interval 95% Confidence Interval
Odds Ratio p‐Value Lower Upper Odds Ratio p‐Value Lower Upper Odds Ratio p‐Value Lower Upper
Category A: Institutional setting versus home/routine care or HHC
Rurality Large fringe metropolitan versus large central metropolitan 0.910 .0005 0.840 0.986 1.011 .0224 0.932 1.096 1.006 .1618 0.933 1.085
Medium metropolitan versus large central metropolitan 0.791 0.707 0.886 0.867 0.769 0.977 0.899 0.798 1.012
Small metropolitan versus large central metropolitan 1.017 0.886 1.167 1.088 0.937 1.263 1.071 0.928 1.235
Micropolitan versus large central metropolitan 0.970 0.860 1.094 1.045 0.921 1.185 0.987 0.874 1.114
Noncore versus large central metropolitan 0.954 0.846 1.075 1.069 0.944 1.210 1.075 0.953 1.214
Sex Female versus male 1.296 <.0001 1.234 1.361 1.364 <.0001 1.299 1.432 1.381 <.0001 1.318 1.446
Race/ethnicity American Indian/Alaska Native versus white 0.925 <.0001 0.661 1.295 0.917 <.0001 0.652 1.289 0.743 <.0001 0.443 1.247
Asian or Pacific Islander versus white 0.744 0.596 0.929 0.613 0.495 0.759 0.609 0.507 0.73
Black versus white 0.695 0.627 0.771 0.765 0.688 0.851 0.816 0.739 0.901
Hispanic versus white 0.680 0.632 0.731 0.663 0.617 0.712 0.679 0.635 0.727
Age group 75–84 years versus 65–74 years 1.759 <.0001 1.663 1.860 1.632 <.0001 1.543 1.725 1.644 <.0001 1.560 1.734
85+ years versus 65–74 years 2.665 2.510 2.830 2.664 2.510 2.828 2.614 2.468 2.768
Risk of mortality Moderate versus minor 1.505 <.0001 1.423 1.592 1.558 <.0001 1.471 1.651 1.489 <.0001 1.410 1.572
Major versus minor 1.820 1.702 1.946 1.900 1.777 2.031 1.892 1.778 2.014
Extreme versus minor 3.586 3.220 3.993 3.547 3.205 3.924 4.213 3.721 4.771
Teaching status Non‐teaching versus teaching 1.414 <.0001 1.239 1.613 1.329 .0003 1.139 1.550 1.261 0.0024 1.086 1.465
Payment source Other versus Medicare 0.426 <.0001 0.356 0.512 0.331 <.0001 0.280 0.391 0.333 <.0001 0.286 0.389
Private versus Medicare 0.731 0.662 0.806 0.704 0.634 0.783 0.730 0.667 0.799
Source of admission Clinical referral versus non‐health care facility 0.661 <.0001 0.601 0.726 0.707 <.0001 0.641 0.780 0.668 <.0001 0.607 0.734
Category B: Institutional setting versus home/routine care
Rurality Large fringe metropolitan versus large central metropolitan 0.884 .0026 0.804 0.972 1.009 .1796 0.916 1.112 1.009 .1315 0.923 1.103
Medium metropolitan versus large central metropolitan 0.790 0.686 0.909 0.864 0.749 0.997 0.950 0.827 1.091
Small metropolitan versus large central metropolitan 0.979 0.825 1.160 1.051 0.878 1.259 1.113 0.941 1.317
Micropolitan versus large central metropolitan 1.019 0.878 1.183 1.018 0.876 1.183 1.030 0.892 1.189
Noncore versus large central metropolitan 0.972 0.839 1.125 1.047 0.902 1.216 1.187 1.025 1.373
Sex Female versus male 1.445 <.0001 1.365 1.530 1.497 <.0001 1.414 1.585 1.559 <.0001 1.476 1.645
Race/ethnicity American Indian/Alaska Native versus white 1.005 <.0001 0.656 1.539 0.936 <.0001 0.625 1.401 0.718 <.0001 0.393 1.314
Asian or Pacific Islander versus white 0.810 0.624 1.052 0.615 0.479 0.790 0.606 0.490 0.750
Black versus white 0.745 0.660 0.842 0.865 0.761 0.983 0.904 0.802 1.019
Hispanic versus white 0.704 0.646 0.767 0.668 0.614 0.727 0.668 0.617 0.723
Age group 75–84 years versus 65–74 years 1.938 <.0001 1.817 2.066 1.737 <.0001 1.629 1.853 1.749 <.0001 1.646 1.859
85+ years versus 65–74 years 3.293 3.066 3.537 3.062 2.852 3.288 3.098 2.892 3.318
Risk of mortality Moderate versus minor 1.623 <.0001 1.520 1.732 1.774 <.0001 1.660 1.896 1.633 <.0001 1.534 1.739
Major versus minor 2.191 2.024 2.372 2.307 2.132 2.495 2.255 2.094 2.427
Extreme versus minor 4.626 4.043 5.294 4.716 4.155 5.353 4.988 4.281 5.811
Teaching status Non‐teaching versus teaching 1.645 <.0001 1.381 1.958 1.556 <.0001 1.288 1.880 1.423 <.0001 1.198 1.690
Payment source Other versus Medicare 0.355 <.0001 0.292 0.431 0.271 <.0001 0.226 0.325 0.271 <.0001 0.230 0.320
Private versus Medicare 0.693 0.619 0.775 0.674 0.598 0.760 0.685 0.618 0.760
Source of admission Clinical referral versus non‐health care facility 0.642 <.0001 0.574 0.717 0.735 <.0001 0.654 0.826 0.677 <.0001 0.605 0.757
Category C: Institutional setting versus HHC
Rurality Large fringe metropolitan versus large central metropolitan 0.936 .0513 0.834 1.051 0.997 .3633 0.888 1.119 1.002 .5348 0.898 1.118
Medium metropolitan versus large central metropolitan 0.807 0.690 0.944 0.918 0.772 1.091 0.858 0.720 1.023
Small metropolitan versus large central metropolitan 1.076 0.884 1.308 1.111 0.899 1.373 1.040 0.841 1.286
Micropolitan versus large central metropolitan 0.890 0.754 1.052 1.101 0.916 1.322 0.943 0.789 1.127
Noncore versus large central metropolitan 0.890 0.752 1.053 1.107 0.926 1.322 0.924 0.776 1.100
Sex Female versus male 1.057 .1327 0.983 1.136 1.151 <.0001 1.073 1.234 1.106 .0039 1.033 1.184
Race/ethnicity American Indian/Alaska Native versus white 0.799 <.0001 0.511 1.249 0.880 <.0001 0.537 1.443 0.834 <.0001 0.380 1.827
Asian or Pacific Islander versus white 0.674 0.493 0.920 0.603 0.449 0.811 0.634 0.488 0.824
Black versus white 0.640 0.554 0.740 0.657 0.570 0.758 0.720 0.628 0.826
Hispanic versus white 0.679 0.613 0.752 0.671 0.606 0.743 0.713 0.647 0.787
Age group 75–84 years versus 65–74 years 1.441 <.0001 1.327 1.565 1.432 <.0001 1.322 1.551 1.439 <.0001 1.332 1.554
85+ years versus 65–74 years 1.862 1.708 2.030 2.089 1.919 2.275 1.967 1.812 2.135
Risk of mortality Moderate versus minor 1.306 <.0001 1.202 1.418 1.245 <.0001 1.144 1.354 1.270 <.0001 1.173 1.375
Major versus minor 1.336 1.213 1.470 1.373 1.248 1.511 1.420 1.299 1.553
Extreme versus minor 2.353 2.019 2.741 2.256 1.957 2.601 3.133 2.604 3.771
Teaching status Non‐teaching versus teaching 1.102 .2868 0.922 1.318 1.035 .7585 0.830 1.291 0.994 .9590 0.793 1.247
Payment source Other versus Medicare 0.731 .0016 0.541 0.988 0.475 <.0001 0.371 0.607 0.549 <.0001 0.432 0.697
Private versus Medicare 0.799 0.690 0.924 0.760 0.650 0.888 0.816 0.712 0.935
Source of admission Clinical referral versus non‐health care facility 0.681 <.0001 0.596 0.779 0.651 <.0001 0.568 0.746 0.614 <.0001 0.539 0.700

Analyses excludes observations coded as HHC discharge.

Analyses excludes observations coded as discharge to home as routine discharge.

Bold indicates significantly different at α = 0.05.

Discharged to HSC Excluding HHC

Table 3 presents results for the likelihood of being discharged to an institutional health care setting versus HSC. Aside from the results of comparisons across race, results of the model with HSC (Table 3) were similar to that of the model using a combined variable for discharge status of HSC or HHC (Table 3). The lower likelihood of being discharged to an institutionalized setting in the previous set of models was not consistent across all years of study for Asian or Pacific Islander individuals when compared to white individuals (no difference detected in 2011) or for black individuals when compared to white individuals (no difference in 2013; Table 3).

Discharged to HHC Excluding HSC

Table 3 presents results for the adjusted analyses for the likelihood of being discharged to an institutional health care setting versus HHC. Aside for comparisons across sex and teaching status, results of the model with HHC (Table 3) were similar to that of the model using a combined variable for discharge status of HSC or HHC (Table 3). There was no difference detected by sex when modeling the likelihood of discharge to an institutionalized health care setting versus HHC in 2011. When modeling the likelihood of being discharged to an institutionalized health care setting versus HHC, we find no evidence to suggest a difference across teaching status for any year under study. Further, variation between the most metropolitan areas and those areas that were classified as large fringe metropolitan was no longer significant for 2011 after controlling for all other terms in the model.

Discussion

The number of older adults suffering a fall‐related hospitalization who were admitted through NIS/CR prior to admission increased from 2011 (n = 41,933) to 2013 (n = 46,056). This is on par with the rise in the older adult Texas population at approximately 10 percent (TDADS 2014), mirroring the growth in fall‐related discharges at approximately 10 percent for the same timeline (2011–2013). However, when assessing the number of overall falls (regardless of location prior to admission) among older adults, we find an increase of 11 percent, slightly higher than the growth in the population. The same comparison from 2007 (n = 42,153; Smith et al. 2010) to 2013 indicates the percent increase at 29.6 percent, higher than the rate of growth in the Texas population aged 65 and older during the same timeline at 25.9 percent (2007, n = 2,346,996; 2013, n = 2,954,614). Thus, the need to identify potential solutions to this issue is growing in terms of the sheer size of those affected by fall‐related hospitalizations.

Limitations

The scope of this study was representative of a large U.S. state effecting generalizability. The specific outcome associated with fall‐related hospitalizations (e.g., hip fracture, cost) or the severity was not identified in the current analyses, but reported elsewhere (Smith et al. 2010; Towne, Ory, and Smith 2014; Towne et al. 2015). Even so, we included risk of mortality to serve as a proxy for severity, which was found to be a strong independent predictor of discharge location/status. In addition, other clinical and patient characteristics (e.g., number of comorbidities) were not available in the data used in this analyses. The inclusion of those with CR may reflect a variety of possible locations prior to admission. Comparisons of the individual characteristics of all fall‐related hospitalizations and all‐purpose hospitalizations of those aged 65 and older were not done as the focus of the study was to compare across discharge status not fall‐related versus the total hospitalized population. Finally, we were unable to measure the discharge status past immediate discharge.

Conclusions and Policy Implications

Several differences persisted even after adjusting for sociodemographic characteristics and a measure of severity (risk of mortality). It is likely that both fragmentation of the health delivery system and patients’ health care needs contribute to the decision to discharge home or to an institutionalized setting. As more baby boomers reach the earlier ages of older adulthood, more may be opting to continue to work and are thereby more likely to be covered on employer‐sponsored health insurance. Thus, identifying potential implications of having Medicare serve as the secondary payer with employer‐sponsored health insurance serving as the primary payer (Goda, Shoven, and Slavov 2007) for a growing population of older adults is timely given different outcomes (i.e., discharge location) across payers identified.

Given that a prior fall is a major risk factor for a recurrent fall regardless of discharge status (Stalenhoef et al. 2002), identifying where individuals go after suffering a fall can help inform where fall‐prevention interventions or fall‐prevention intervention referrals may be delivered most effectively. Improvements in discharge planning for older adults suffering fall‐related hospitalizations have been suggested as a potential strategy in the prevention of recurrent falls (Lim, Hoffmann, and Brasel 2007).

Discharge planning is a critical part of care with the goal of reducing unexpected readmissions and improving communication within care transitions (Shepperd et al. 2013). The discharge process may include prescription medication management and follow‐up medical care (Spinewine et al. 2013) in addition to related self‐care planning. The fragmentation of the health care system may make this process more challenging in that pharmacies, primary care settings, post‐acute care settings, and hospitals may not exchange information or coordinate service delivery between settings. Thus, the patient is left with the task of communicating information between care settings. This is a task which is not well suited to older adults following a fall‐related hospitalization. Evidence suggests that tailored or individualized discharged planning for older adults may be associated with reduced hospital readmissions (Shepperd et al. 2013) and is therefore critical. Policies affecting what is included in standardized discharge planning for this subpopulation should include three essential components: evidence‐based fall‐prevention programs, community‐based fall‐prevention programs, and information on how to access those programs.

Supporting information

Appendix SA1: Author Matrix.

Acknowledgments

Joint Acknowledgment/Disclosure Statement: We would like to acknowledge these affiliated institutions: Texas A&M University; the University of Georgia; the University of Memphis; University of Texas Southwestern Medical Center; and the University of Arkansas for Medical Sciences as listed on our respective affiliations.

All authors declare no relevant financial interests, activities, relationships, and affiliations (other than those affiliations listed in the title page of the manuscript) including, but not limited to, employment, affiliation, funding, and grants received or pending, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Disclosures: None.

Declaimer: None.

References

  1. Bell, B. A. , Ene M., and Schoeneberger J.. 2013. “A Multilevel Model Primer Using SAS PROC MIXED” SAS Global Forum: Statistics and Data Analysis. Conference Proceedings, Paper 433‐2013, pp. 0–19.
  2. Das, A. , Norton E. C., Miller D. C., and Chen L. M.. 2016. “Spending in the Post‐Acute Care Setting Accounts for a Sizeable Portion of Spending Associated with a Hospitalization Episode.” JAMA Internal Medicine 176 (1): 117. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Goda, G. S. , Shoven J., and Slavov S. N.. 2007. “A Tax on Work for the Elderly: Medicare as a Secondary Payer.” National Bureau of Economic Research. NBER Working Paper No. 13383, pp. 1–25.
  4. Lim, H. J. , Hoffmann R., and Brasel K.. 2007. “Factors Influencing Discharge Location after Hospitalization Resulting from a Traumatic Fall among Older Persons.” Journal of Trauma and Acute Care Surgery 63 (4): 902–7. [DOI] [PubMed] [Google Scholar]
  5. Masud, T. , and Morris R. O.. 2001. “Epidemiology, of Falls.” Age and Ageing 30: 3–7. [DOI] [PubMed] [Google Scholar]
  6. Mor, V. , Rahman M., and McHugh J.. 2016. “Accountability of Hospitals for Medicare Beneficiaries’ Postacute Care Discharge Disposition.” JAMA Internal Medicine 176 (1): 119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Rice, N. , and Leyland A.. 1996. “Multilevel Models: Applications to Health Data.” Journal of Health Services Research 1 (3): 154–64. [DOI] [PubMed] [Google Scholar]
  8. Shepperd, S. , Lannin N. A., Clemson L. M., McCluskey A., Cameron I. D., and Barras S. L.. 2013. “Discharge Planning from Hospital to Home.” Cochrane Database of Systematic Reviews 1 (1). [DOI] [PubMed] [Google Scholar]
  9. Smith, M. , Ory M., Beasley C., Johnson K., Wernicke M., and Parrish R.. 2010. “Falls among Older Adults in Texas: Profile from 2007 Hospital Discharge Data.” TPHA Journal 62 (1): 15–20. [Google Scholar]
  10. Spinewine, A. , Claeys C., Foulon V., and Chevalier P.. 2013. “Approaches for Improving Continuity of Care in Medication Management: A Systematic Review.” International Journal for Quality in Health Care 25 (4): 403–17. [DOI] [PubMed] [Google Scholar]
  11. Stalenhoef, P. , Diederiks J., Knottnerus J., Kester A., and Crebolder H.. 2002. “A Risk Model for the Prediction of Recurrent Falls in Community‐Dwelling Elderly: A Prospective Cohort Study.” Journal of Clinical Epidemiology 55 (11): 1088–94. [DOI] [PubMed] [Google Scholar]
  12. Sterling, D. A. , O'Connor J. A., and Bonadies J.. 2001. “Geriatric Falls: Injury Severity Is High and Disproportionate to Mechanism.” Journal of Trauma and Acute Care Surgery 50 (1): 116–9. [DOI] [PubMed] [Google Scholar]
  13. TDADS . 2014. “Texas Department of State Health Services Census Data“ [accessed on June 06, 2014]. Available at http://www.dshs.state.tx.us/chs/popdat/detailX.shtm
  14. THHS . 2011. Texas Hospital Inpatient Discharge Public Use Data File. Austin, TX: Texas Health and Human Services, Center for Health Statistics. [Google Scholar]
  15. Tinetti, M. E. , and Williams C. S.. 1997. “Falls, Injuries Due to Falls, and the Risk of Admission to a Nursing Home.” New England Journal of Medicine 337 (18): 1279–84. [DOI] [PubMed] [Google Scholar]
  16. Towne, S. , Ory M., and Smith M.. 2014. “Cost of Fall‐Related Hospitalizations among Older Adults: Environmental Comparisons from the 2011 Texas Hospital Inpatient Discharge Data.” Population Health Management 17 (6): 351–6. [DOI] [PubMed] [Google Scholar]
  17. Towne, S. D. , Smith M. L., Yoshikawa A., and Ory M. G.. 2015. “Geospatial Distribution of Fall‐Related Hospitalization Incidence in Texas.” Journal of Safety Research 53: 11–6. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Appendix SA1: Author Matrix.


Articles from Health Services Research are provided here courtesy of Health Research & Educational Trust

RESOURCES