“Preventable hospitalizations” in older adults with dementia: Are they really preventable?

INTRODUCTION

The concept of preventable hospitalization originated to characterize hospital admissions resulting from lack of timely access to primary care. Initially a measure of access to primary care among marginalized communities,¹ the concept has since become ingrained as a quality of care measure, including for persons with dementia.²

An analysis of preventable hospitalizations within the Adult Changes in Thought (ACT) cohort found that, among those with and without a research-derived diagnosis of dementia, rates were significantly higher among those with dementia,³ consistent with other analyses.^4,5 Yet why dementia might be associated with increased risk of admission is unclear.

To begin to examine these questions, we conducted a retrospective review of medical records from a subset of ACT participants who experienced preventable hospitalization. We focused on admissions for congestive heart failure (CHF) or bacterial pneumonia (BP), the most common chronic and acute preventable admission conditions.³

METHODS

Participants were drawn from the ACT cohort, a population-based study of aging and dementia risk.⁶ We identified those hospitalized for CHF or BP on/after January 1, 2006 using principal discharge diagnosis International Classification of Disease codes. For each diagnosis, a random sample of 15 participants with ACT-diagnosed dementia and 15 age-matched participants without dementia were selected, yielding a total sample of 60.

We conducted a chart review (Supplementary Text S1) of ambulatory, urgent, emergency department, and admission records to inform three questions: 1) Could admission have been prevented through more timely receipt of care? 2) Among those with dementia, did dementia elevate admission risk and, if so, how? 3) Among those with dementia, was admission risk moderated by whether clinicians recognized cognitive impairment? Charts were independently reviewed by the authors with discussion to resolve differences.

Group Health Research Institute and Michigan Medicine IRBs approved the study.

RESULTS

Two of the sample of 60 lacked outpatient records, yielding a final sample of 29 CHF (14 without, 15 with dementia) and 29 BP (15 without, 14 with dementia) admissions. Mean age was 88.3 years (standard deviation 7.5); 58.6% were female (Table 1).

Table 1.

Characteristics of Adult Changes in Thought Participants Without and With Dementia Who Experienced Preventable Hospitalization

	Congestive Heart Failure		Bacterial Pneumonia
Characteristic, n (%; 95% CI)	Without dementia n=14	With dementia n=15	Without dementia n=15	With dementia n=14
Age, y; mean (SD; 95% CI)	89.4 (7.0; 85.4, 93.4)	89.6 (6.1; 86.2, 93.0)	86.5 (8.6; 81.7, 91.3)	87.6 (9.0; 82.4, 92.8)
Female	8 (57.1; 28.9, 82.3)	9 (60.0; 32.3, 83.7)	9 (60.0; 32.3, 83.7)	8 (57.1; 28.9, 82.3)
Delay in care	2 (14.3; 1.8, 42.8)	3 (20.0; 4.3, 48.1)	6 (40.0; 16.3, 67.7)	4 (28.6; 8.4,58.1)
Presentation to primary carea	0 (0; 0, 23.2d)	3 (20.0; 4.3, 48.1)	3 (20.0; 4.3, 48.1)	2 (14.3; 1.8, 42.8)
Primary care deliveryb	2 (14.3; 1.8, 42.8)	0 (0; 0, 21.8d)	3 (20.0; 4.3, 48.1)	2 (14.3; 1.8, 42.8)
Dementia diagnosis presentc	n/a	5 (33.3; 11.8, 61.6)	n/a	10 (71.4; 41.9, 91.6)

CI: 95% confidence interval

Note: 95% confidence intervals for binary data are Binomial exact confidence intervals.

^aReceipt of care was delayed by a gap between onset of symptoms (as documented) and when the patient presented for medical attention; upon presentation, the care provided was appropriate.

^bThere was no delay documented between symptom onset and presentation; rather, there was a gap between presentation and subsequent delivery of appropriate clinical care.

^cThose with a dementia diagnosis documented by clinicians among those with dementia as determined by the ACT study.

^dOne sided 97.5% Binomial exact confidence interval.

Delayed care was uncommon (15/58 [25.9%; 95% confidence interval (CI) 15.3%, 39.0%]); representative cases are presented in Table 2). Among those without dementia, evidence from eight of 29 admissions (8/29 [27.6%; CI 12.7%, 47.2%]) suggested that ambulatory care was delayed. Among those with dementia, a similar proportion experienced delay (7/29 [24.1%; CI 10.3%, 43.5%]).

Table 2.

Select Instances of Potentially Delayed Care

Without dementia
Pneumonia	73y, F Presented to primary care with a cough for 3 days, “similar to prior cough illness thought to be pneumonia” Prescribed an oral antibiotic with plan to “fill later in case of worsening”; no additional evaluation or follow-up scheduled Several days later, she was admitted to the hospital with pneumonia and sepsis
Congestive Heart Failure	93y, F Contacted her outpatient team, got initial recommendations by telephone (e.g., elevating bed for comfort) No further evaluation or outpatient visit scheduled Friend visited several days later, found her in severe respiratory dress, called 911 on her behalf
With dementia
Pneumonia	76y, M Patient presents to urgent care with sore throat and headache Seen the next day in clinic with occasional temperature to 102F and confusion; discharged with plan to take acetaminophen Patient on multiple chronic medications, will not accept help or reminders from his wife. The third day, after multiple calls by wife to nurse triage plus another urgent care visit prompted by fever and confusion (e.g., “peeing on the floor”), he was admitted.
Congestive Heart Failure	81y, F Patient presented to the outpatient office reporting worsening symptoms (e.g., productive barking cough) over 3 days; symptom onset had been 10 days earlier “grunting respirations, speaks in short sentences” Documentation mentions spouse with “advanced dementia” Presented with pulse oximetry on room air 88%, respiratory rate of 30/min 911 called by clinic staff

Patients or families introduced care delays through time lapse between symptom onset and contacting the clinician. For example, an 80-year-old (with dementia) was brought to see her primary care physician (PCP) by her daughter; with “cough and general malaise over the last few days … unable to get up and go to the dining room …” There was no prior contact with the PCP for these symptoms.

There was no apparent mechanism whereby dementia contributed to risk of preventable hospitalization. Recognition of dementia (i.e., a diagnosis present in clinical documentation) did not appear to contribute to delayed care or otherwise increase admission risk.

DISCUSSION

In this preliminary study, we found that about one-quarter of hospitalizations for CHF or BP may have been preventable via earlier receipt of medical care. Among these, delays were as often attributable to older adults and their care supports (family or congregate living facility) as to their clinicians. Preventability was similar regardless of dementia status. Among those with (research-diagnosed) dementia, care delays did not seem any more likely among those whose dementia was undocumented in the medical record.

While a majority of hospitalizations reviewed did not appear preventable, one quarter may have been. Efforts to reduce hospitalizations of persons with dementia, whether preventable or all-cause, could include “do not hospitalize” orders, carer education/support related to the decision to hospitalize;⁷ wider availability of “hospital-at-home”;⁸ hospital outreach services to optimize admissions for those at risk⁹; or a Clinical Decision Unit that applies alternatives to hospitalization.¹⁰

This study has limitations. The sample focused on two conditions; our findings may not apply to other preventable admission conditions. It was conducted within a single managed care organization in one region of the United States, so results may not be generalizable to other settings. The study relied on clinicians’ documentation: if factors were undocumented, it is unclear whether the clinician was aware of a particular factor (e.g., dementia) and did not consider it relevant or if the factor was present but the clinician did not recognize it.

In conclusion, this analysis suggests that relatively few “preventable” hospitalizations of older adults—at least for the conditions considered here—may actually be preventable.

Supplementary Material

Supinfo

Supplementary Text S1. Preventable hospitalization chart abstraction form.