Descriptive Analysis
Tables 1 and 1a include descriptive statistics. The analytical file included data on 110 patients. The typical enrollee is a 60-year old single African-American male living with two other individuals. He has a PCP and takes him 23 minutes to get to his/her office, and has Medicare coverage but is not dually eligible. He takes 12 medications for at least one chronic condition and is compliant with his treatment.
Eighty-four (76%) patients were 49 years or older and 81 (74%) were African-American, with an almost even split between men and women. Sixty-nine were single (68%), 25 (25%) were married and 104 (95%) were introduced to MIH via phone call. Sixty-five (74%) had low fall risk scores with an average score of 2.5. One-hundred (91%) were sometimes or always compliant with their medication regimens and took an average of 12 medications, with 77 (76%) of them taking 10 or more medications. The majority (102, 93%) have a PCP and all 110 patients have insurance, with Medicare, Medicaid and private insurance covering 61 (56%), 35 (32%) and 14 (13%) patients, respectively. Patients took an average of 23 minutes to get to their PCP, with 101 (82%) taking anywhere from 10 to 39 minutes. Seventy-five (68%) patients had hypertension followed by 46 (42%) with diabetes, 40 (36%) with psychiatric/behavioral disorders, 37 (34%) with asthma/COPD, 29 (26%) with hypercholesterolemia, 26 (24%) with Chronic Heart Failure (CHF) and 25 (23%) have had a stroke/CVA.
[Tables 1 and 1a here]
From the 110 patients in the full sample, 45 (41%) were classified as High-Frequency (HF) users, with the other 65 (69%) classified as Low-Frequency (LF) users of EMS services (Tables 1 and 2). The HF group was relatively younger, 58 years on average, and had a larger proportion of women (26, 58%). They also had slightly higher fall risk scores (2.8) and were less compliant with medications. In terms of access to care, the rates for PCP were higher on the HF group, where 42 (93%) had a PCP compared with 60 (92%) on the LF group. Dual eligibility was higher on the HF group with 10 (22%) patients, compared with the LF group where only eight (12%) patients. The HF group had lower rates of private insurance and a slightly larger share of Medicaid beneficiaries. Travel times to PCPs were similar between groups. Asthma/COPD and hypercholesterolemia were slightly more prevalent in the HF group, although they had higher rates of CHF and psychiatric/behavioral illness. Hypertension and diabetes were not as prevalent in the HF group as with the LF group.
Since a large number of MIH patients (65) were not HF users of EMS services, non-parametric statistics were run to determine if there were any associations between covariates in the HF and LF groups that could influence 911 calls and transports counts. Based on a 95% significance level, the study found no statistically significant associations between sociodemographic, assessment & medications, insurance/access to care or clinical/chronic illness covariates and eligibility, with only moderate, not statistically significant associations between persons living in residence, compliance with medications, dual eligibility, CHF and psychiatric/behavioral illnesses. Given these results, it was appropriate to use the full sample to test for predictors of EMS calls and transports. Prior to running the regression, collinearity tests revealed no significant correlations between covariates.
Regression model
Table 2 and 2a include the regression results for EMS calls and transports. For EMS calls, patients ages 65 and older were less likely to call EMS compared to their younger counterparts. Patients ages 64 to 78 were 49% (p-.005) less likely to call EMS, whereas those 79 years and older were 92% (p=.000) less likely to call EMS, compared to the youngest group. Patients who were either married or divorced were 65% (p=.000) and 51% (p=.001) less likely to call EMS, respectively, compared to single patients. In contrast, widowed patients were more than 4 times more likely to call EMS (p=.003). Men were two (2) times (p=.000) more likely to call EMS compared to women. In terms of race, Blacks had 89% (p=.000) higher chance to call EMS and Hispanics showed a 7-fold increase (p=.000) in their chance of calling EMS, when compared to White patients. Patients with a high fall risk score were 2.7 times (p=.000) more likely to call EMS compared to patients with low scores. Patients living with three to six additional people in the same residence had more than twice the chance of calling EMS compared to patients living with one person.
Patients with asthma/COPD were 2.7 (p=.000) times more likely to call EMS, and those with hypertension, CVA/Stroke, and psychiatric or behavioral conditions had between 48% and 58% chance of calling EMS compared with patients without these illnesses. Patients with high cholesterol were 33% less likely to call EMS and those with diabetes were 10% more likely to call EMS compared with patients without these illnesses, but the results were not statistically significant. Compliance with medications showed an increased risk for calling EMS, but the results were not statistically significant. For each medication patients took, the changes of calling EMS increased by 4% (p=.000). Patients covered through Medicaid were 71% (p=.000) less likely to call EMS compared with patients with private insurance coverage, whereas patients eligible for dual coverage were 49% (p=.000) less likely to call EMS. Travel times to PCP offices showed large and statistically significant results. Patients travel times greater than 30 minutes were between 10 and 17 times (p=.000) more likely to call 911 compared with patients travelling less than 10 minutes to their PCP office.
[Tables 2 and 2a here]
Patients 49-64 and 65-78 years were 46% (p=.02) and 48% (p=.04) less likely to be transported compared to those ages 19-33, whereas patients 79 and older were 91% (p=.000) less likely to be transported compared to the youngest group. Marital status had a similar effect as 911 calls. Married patients were 72% (p=.000) less likely to require transport compared to single patients. Divorced patients also showed a protective effect, with 56% (p=.003) less chance for transport. In contrast, widowed patients were 10 times (p=.000) more likely to require transport compared with single patients.
Males were 2.4 times (p=.000) more likely to be transported compared to women. Blacks and Hispanics were 1.9 (p=.002) and 3.6 (p=.000) times more likely to require transport, respectively, compared with Whites. Living with three to six people in the same residence increased the chances of transport anywhere between 1.6 and 4.9 times, compared with living with one person only.
Chronic illnesses differed somehow between transports and calls. Patients with diabetes, psychiatric or behavioral illnesses, CVA/Stroke, high cholesterol and asthma/COPD were more likely to require transport compared with patients without any of these conditions. Asthma/COPD, psychiatric/behavioral conditions and diabetes had the largest effect, with 4.3 (p=.000), 1.8 (p=.000) and 1.9 (p=.000) higher chances of transport, respectively, compared with patients without these conditions. Patients who were sometimes compliant with their medications were 50% (p=.008) more likely to require transport compared with patients who were never compliant. The effect on the number of medications was similar as with 911 calls, with a 4% (p=.000) chance of transport for each medication. Patients with Medicare coverage were 2.1 times (p=.002) more likely to require transport compared with those with private insurance. Travel times for 911 transports were significant and showed increased likelihood of transport by up to 15 times (p=.000) when travel exceeded 40 minutes or longer, compared to travel times shorter than 10 minutes.
Sensitivity analysis
After dropping non-significant variables from the model, the likelihood of EMS calls remained for patients who were married and widowed, with a high fall risk score, patients who referred CVA, psychiatric or behavioral illness, CHF and asthma/COPD, the number of prescription medications, and travel times. For EMS transports, the effects remained for married or widowed male patients, those with asthma/COPD, CHF, psychiatric/behavioral illnesses, the number of prescription medications and travel times.
911 call and transport data
As shown in Table 3 and figure 1, both calls and transports experienced sharp reductions 30 days after the first patient visit by the MIH team, 75% and 79%, respectively. At the 4th month mark, calls and transports remained 11% and 16% below baseline although both increased compared to the 30-day mark. By the 6th month mark, both call and transports increased by 24% and 18%, respectively, compared with the baseline.
[Table 3]