For more than 20 years organizations and policy makers looked for cost estimates associated with loneliness and social isolation. This summer a client wanted a sense of those costs. Using recent literature I derived a rough estimate of health care expenditures linked to loneliness: about 2.5% of market-based medical costs for seniors. The background and methods are given below the fold.
An Initial Guess
It’s sound practice before beginning a quest for a point estimate to have an approximate idea of the answer; if nothing more, at least a guess. I knew from a review of the literature that dementia, mainly a senior’s disease, accounts for as much as a quarter of senior health care expenditures (HCE).1 Surely loneliness-related HCE would be considerably less than that. I guessed it to be no greater than 5%.
History and Scope
In 1992 the Institute of Medicine published a report, “The Second Fifty Years,” which notes that the cost burden of social isolation is an area2
…which requires the attention of carefully designed and scientifically managed research. p247
Fast forwarding 23 years to 2014, from a report by the British Campaign to End Loneliness, the research’s original intent was3
To establish what the financial costs of loneliness are to the state, and to wider society; in particular, focusing on costs that health and well-being commissioners prioritize in their decision making. p1
And still there had been no cost estimate.
My impetus to estimate costs was a small project for Empire Health Foundation, Spokane, WA. I don’t claim “carefully designed and scientifically managed research,” but simply a starting value for immediate use with the hope there will be subsequent improvement, perhaps by funded researchers. The focus was on formal direct health care costs, those purchased in the market place. Not considered were additional costs such as family caregiving, multiplier effects, or opportunity costs like lost income.
Background on National HCE
According to David Lassman et al.,4 total National Health Care Expenditures in 2010 was $2.2 trillion. That’s direct costs to treat or prevent morbidity. Of that $2.2T, one-third (34%) or $744B was spent on health care for seniors ($18,424 per person aged 65 plus). Note that seniors account for 34% of costs but were 13% of the population.
However those costs are not distributed evenly. In 2010, ten percent of seniors accounted for half (50.1%) of senior HCE or almost $50K ($49,639) per person.5 During the analysis I thought of that as a 10-50-50 rule. I bring this up because it helped me identify the information I then used to derive the 2.5% estimate. The trick in an endeavor such as this is finding the pieces that fit together, to unify, which is seldom obvious in the beginning. The 10-50-50 rule helped as a simple model for context.
Further Background: on Loneliness and Dementia among the Elderly
Loneliness and social isolation can have many bad effects, some of which were noted in the previous post. Furthermore loneliness and social isolation are not necessarily the same thing. As an example consider this excerpt from an essay by Stephen Fry:
Lonely? I get invitation cards through the post almost every day. I shall be in the Royal Box at Wimbledon and I have serious and generous offers from friends asking me to join them in the South of France, Italy, Sicily, South Africa, British Columbia and America this summer. I have two months to start a book before I go off to Broadway for a run of Twelfth Night there.
I can read back that last sentence and see that, bipolar or not, if I’m under treatment and not actually depressed, what the f̶̶ ̶ ̶ right do I have to be lonely, unhappy or forlorn? I don’t have the right. But there again I don’t have the right not to have those feelings. Feelings are not something to which one does or does not have rights.
In the end loneliness is the most terrible and contradictory of my problems.
About the deleterious effects of loneliness, if you were to do an internet search on loneliness health effects or similar terms, you’d find plenty of material. And it gets worse as you get older. Nor is it clear what to do about loneliness. Interventions for social isolation are easier to implement, but it’s loneliness that, besides being less discernible, is the worse. Loneliness, feeling empty or desolate, for the elderly is linked with significant decline in daily activities of living, increased risk of early death (about the same risk as disadvantaged socioeconomic status), twice as unhealthy as obesity, and among other bad effects, an increased risk for dementia. (Note that Alzheimer’s constitute 60%-90% of dementia, estimates vary.)
The Amsterdam Study
A prospective study of almost 2,200 community-living seniors in Amsterdam looked at risk factors for dementia over a three-year period.6 (So the results may not apply to rural populations, even if they do apply to other large thriving cities.) Among the interesting results from this study was that 19.9% of the sample self-identified as feeling lonely or very lonely. Furthermore, this group had an odds ratio of 1.64 of developing clinically diagnosed dementia over three years. Using a now standard method for converting the odds ratio to a relative risk which is more easily interpretable, we get a small reduction from an odds ratio of 1.64 to a relative risk of 1.58, which is what I used.7 The loneliness prevalence and increased risk of dementia, along with related costs for the senior U.S. population, were the input for the 2.5% estimate.
The Amsterdam study is the only one I know that gives significant estimates of apparently dementia-free but lonely seniors who a few years later were diagnosed with dementia. For robust results we need replication but for now I believe this is all we’ve got. Of course, we could always wait another 23 years during which time we might write about how we need a loneliness cost estimate.
For this starting value, we have new and helpful information like the Amsterdam study and a recent Health Affairs article on HCE by age groups. Without stepping through it, however, I suspect one could have gotten a handle on loneliness costs earlier via data links between loneliness, depression and dementia, and increased falls associated with dementia.
Deriving the Estimate
There’s considerable uncertainty in the literature regarding the prevalence of dementia as well as its cost. For example, Hurd et al.8 give a total 2010 HCE estimate (including the costs of coexisting conditions) of $109B for all ages 18 and above, while the Alzheimer’s Association estimate for similar existing conditions but for seniors only is $172B. (Regarding ages 18 plus, about 10% of all dementias occur prior to ager 65.) Further contributing to the variability, dementia prevalence doubles roughly every five years of age from age 70 on, while recent research indicates the incidence of dementia is decreasing, so dementia cost estimates are a moving target.9 If you take the Hurd et al. per capita estimate of $33.3K for dementia with comorbidities and multiply it by 40M (U.S. 2010 senior population count) times 6.2%, a recent estimate, you get $83B instead of Hurd’s $109B. The Alzheimer’s Association claims 11% of seniors 65 and older have Alzheimer’s disease and that in 2010 dementia cost $172B. Those are big differences, $83B, $109B and, $172B, all supposedly measuring close to the same thing. (The range is $89B, which is more than double the low estimate.)
I assume that both 19.9% of seniors feel lonely or very lonely and that a 58% increased risk for the lonely to develop dementia are representative. That suggests loneliness contributes an additional 11.6% cases of dementia (19.9% X 0.58 = 11.6%). That was in Denmark which I now apply to U.S. data. Taking the extreme estimates from above to estimate the percentage of senior HCE:
11.6% X $83B = $9.6B, or as a percentage $9.6B is 1.3% of $744B
11.6% X $172B = $20B, or 2.7% of $744B.
This denotes a cost range for the loneliness that leads to dementia. However loneliness is a risk factor for more than just dementia. Unfortunately I don’t have loneliness-related estimates for those additional morbidities so I make a guess: I add 0.5% to the above percentages to arrive at a more inclusive 1.8% to 3.2%. And for a single point estimate I use the average, 2.5%, as a first cut at the percentage of HCE attributable to loneliness.
To put this point estimate in context, note that 2.5% of $744B (total senior HCE in 2010) is $18.6B, which is roughly half of the Medical Expenses Panel Survey estimate of $34.5B for senior cancer treatment in 2012. (The two most expensive conditions for seniors are heart conditions followed by cancers.)10
As already noted there’s large variability of prevalence and cost estimates of dementia. And the 2.5% loneliness estimate is hardly robust, depending as it does on a single study from another continent.
The Alzheimer’s Association 2014 report11 touches on new research regarding pre-clinical Alzheimer’s. Assuming the pre-clinical research eventually bears out, it’s possible that the Amsterdam study subjects who were identified as lonely with no clinical evidence of dementia may already have had incipient Alzheimer’s.
My intuition is Hurd et al. underestimates the costs associated with coexisting conditions. I included comorbidities because they are intrinsically part of the David Lassman et al. costs estimates, and did not have reason to separate them out. Hurd et al. did separate them and so had two per capita estimates, $33.3K with coexisting conditions, and $28.5K without and other adjustments. That $4.8K difference seems too low. There’s much literature on the higher prevalence of comorbidities associated with dementia. Take falls as an example. Dementia patients fall about two-three times as often (as I recall) as those not mentally impaired, increasing their likelihood for trauma and hospitalization. An average hospital stay for a fall is over $30K, and most of those patients then go on to a nursing home for months (or the rest of their life). Nursing home expenses are over $80K per year.
I focused on loneliness, not social isolation. Obviously the two often go together, it’s just that they are not necessarily the same: a person could be lonely without being socially isolated (some marriages, for example) or be socially isolated and not be lonely (for instance, embracing solitude). The Amsterdam study looked at social isolation but did not find a significant relationship with dementia. However, we know from other studies that social isolation adversely affects health (e.g., increased prevalence of decline in activities in daily living) so it has a health care cost. At this point I can only guess at what that might be. My sense from the literature is loneliness is the far worse of the two. As a guess I’d think social isolation as a separate factor adds another 1%, so together, loneliness and social isolation might amount to 3.5% of total senior HCE.
Here I’m sharing a health care cost estimate for loneliness in the spirit that it might be useful and perhaps a spur for an improved estimate. In the meantime, 2.5% seems reasonable. Earlier I noted the 10-50-50 rule, 10% of the senior population accounted for 50% of total senior HCE or $50K. That rule helped lead me to using the Amsterdam study and dementia for a first cut at loneliness costs. One reason it helped was there are many moving parts throughout the loneliness and dementia literature and little standardization (e.g., some studies are about Alzheimer’s, others about dementia; ages ranged between 18-plus to 71-plus). There were times early on I was getting confused enough that I wondered if I were getting dementia. Having that 10-50-50 anchor, a simple model I could keep at hand, helped to keep me centered during the investigation.12
For the future, conceptually there’s an easy improved solution. Under a single-payer system or an HMO, sample the in-house population, identify those lonely from those not, follow them for, say 2-4 years, at the end of which—making sure to identify those who are lonely for the duration or only part way—average the HCE for each group. Then you’d have an estimate of the health care costs associated with loneliness.
- The Alzheimer’s Association projected $172B HCE for dementias in 2010 or 23% of total senior HCE of $744B, from 2010 Alzheimer’s Disease Facts and Figures, p34. ↩
- http://www.nap.edu/catalog.php?record_id=1578 ↩
- http://www.campaigntoendloneliness.org/wp-content/plugins/download-monitor/download.php?id=181 ↩
- David Lassman et al., “US Health Spending Trends By Age And Gender: Selected Years 2002-10,” Health Affairs, 33, no.5 (2014): 815-822. This report is essentially by the Office of the Actuary, Centers for Medicare and Medicaid Services. ↩
- Steven B. Cohen and Namrata Uberoi, “Statistical Brief #421: Differentials in the Concentration in the Level of Health Expenditures across Population Subgroups in the U.S., 2010,” MEPS/AHRQ (August 2013), http://meps.ahrq.gov/data_files/publications/st421/stat421.shtml. ↩
- Tjalling Jan Holwerda et al., “Feelings of loneliness, but not social isolation, predict dementia onset: results from the Amsterdam Study of the Elderly (AMSTEL),” J Neurol Neurosurg Psychiatry doi:10.1136/jnnp-2012-302755v1 ↩
- jama.jamanetwork.com/article.aspx?articleid=188182 ↩
- Michael D. Hurd et al., “Monetary Costs of Dementia in the United States,” N Engl J Med (2013) 368:1326-1334, ↩
- http://www.nytimes.com/2013/07/17/health/study-finds-dip-in-dementia-rates.html?_r=0 ↩
- http://meps.ahrq.gov/mepsweb/data_stats/tables_compendia_hh_interactive.jsp?File=HCFY2012&Table=HCFY2012_CNDXP_C&TCOPT1=&TCOPT2=&ModPopCheck=Yes&ModPop=Yes&_Debug=&DDAGELOW=65&DDAGEHIGH=100&Action.x=0&Action.y=0 ↩
- I used two of their Facts and Figures reports, http://www.alz.org/downloads/Facts_Figures_2014.pdf and http://www.alz.org/documents_custom/report_alzfactsfigures2010.pdf. ↩
- Which relates to one of my very favorite quotes: To understand is above all to unify, by Albert Camus in “The Myth of Sisyphus and Other Essays.” ↩