Costs of patients hospitalized for multidrug-resistant tuberculosis

SUMMARY

SETTING

From 1993 through 1998, 1846 cases of multidrug-resistant tuberculosis (MDR-TB) were reported in the United States. Costs associated with MDR-TB are likely to be much higher than for drug-susceptible tuberculosis due to longer hospitalization, longer treatment with more expensive and toxic medications, greater productivity losses, and higher mortality.

OBJECTIVE

To measure the societal costs of patients hospitalized for MDR-TB.

DESIGN

We detailed in-patient costs for 13 multidrug-resistant patients enrolled in a national study. We estimated costs for physician care, out-patient treatment, and productivity losses for survivors and for deceased patients.

RESULTS

In-patient costs averaged US$25 853 per person and $1036 per person-day of hospitalization. Out-patient costs per person ranged from $5744 to $41821 (average $19028, or $44 a day). Direct medical costs averaged $44 881; indirect costs for those who survived averaged $32 964, and indirect costs for those who died averaged $686 381 per person. Total costs per person ranged from $28 217 to $181 492 (average $89 594) for those who survived, and from $509 490 to $1 278 066 (average $717 555) for those who died.

CONCLUSION

The societal costs of MDR-TB varied, mostly because of length of therapy (including in-patient), and deaths during treatment.

Active tuberculosis (TB) disease is generally curable with a 6-month regimen that includes isoniazid (INH) and rifampin (RMP), with pyrazinamide (PZA) and ethambutol (EMB) during the first 2 months.¹ Drug-resistant TB is also usually curable, but requires longer treatment with different and less effective medications.¹

The Centers for Disease Control and Prevention (CDC) have been routinely monitoring drug-resistant TB through the national TB surveillance system since 1993. A total of 1846 multidrug-resistant tuberculosis (MDR-TB) patients were reported in the United States from 1993 through 1998; approximately 31% of MDR-TB patients die during therapy, and 4% are diagnosed at death (M Moore, CDC, personal communication, 2003). MDR-TB currently accounts for approximately 1% of US TB cases.

Past studies of MDR-TB costs have used different methods and arrived at a wide range of estimates. In-patient treatment charges for MDR-TB were first estimated at US$180 000 in 1990 dollars by Mahmoudi and Iseman.² Palmer et al. estimated average MDR-TB hospitalization charges from a 16-state hospital discharge database for immunocompetent and immunosuppressed patients in 1992 dollars at $21 966 and $76 881, respectively.³ Burman et al. used Mahmoudi and Iseman’s figure, converted it to a cost (as costs are generally about one half of charges), and updated it to 1994 dollars to estimate the average in-patient cost of treating MDR-TB at $115 740. Cost estimates for diagnosis, out-patient directly observed therapy (DOT), and patient productivity losses were then added for 3 years of treatment, discounting the values for years 2 and 3 at 5% annually, for a total cost of $146 271 in 1994 dollars.⁴

Various costs associated with MDR-TB are much higher than those for drug-susceptible TB due to longer hospitalization, longer and more complex treatment with more expensive and toxic medications, greater productivity losses, and higher mortality. The objective of this study was to estimate the cost of patients hospitalized for MDR-TB in the US, detailing direct medical in-patient and out-patient costs and indirect costs. The value of these estimates stems from the use of a multi-site cohort of patients. They will also assist in the evaluation of net benefits of interventions to prevent MDR-TB.

STUDY POPULATION AND METHODS

We analysed in detail the costs for 13 MDR-TB patients who were part of a cohort of 1493 TB patients reported to the CDC by 10 US TB programs (from California, Georgia, Illinois, Mississippi, New York, South Carolina, and Texas) during 6 months in 1995/1996 and enrolled in a CDC study of factors and costs associated with TB hospitalization. Of the total cohort, 733 patients were hospitalized for TB (defined as TB diagnosis during hospitalization or receipt of primary discharge diagnosis of TB) and costs collected on their in-patient care. The methods of the original study are described elsewhere.⁵ The CDC’s Institutional Review Board approved the original study with written informed consent from each patient.

In the original cohort of 1493 TB patients, there were 24 MDR-TB patients, 17 of whom were hospitalized for TB. It was possible to disaggregate hospitalization costs into various categories for only 13 of them; the other four only had aggregated costs for their entire hospitalizations. For these 13 MDR-TB patients, we also estimated other costs that were not collected as part of the original study, including physician costs, out-patient medication and DOT costs, productivity losses for the survivors, and the value of lives lost for the deceased using standard cost-of-illness methods.⁶ We did not measure or estimate any incentives or enablers that may have been part of treatment.

We grouped the patients into three illness severity categories: severe (those who died during therapy), moderate (those who were human immunodeficiency virus [HIV] infected, had cavitary disease, or had TB surgery), and mild (all others). This categorization scheme was derived through consultation with a physician who cares for MDR-TB patients and seemed to correlate well with therapy length, which is also an indicator of disease severity for those who survived. As reporting of resistance to second-line TB medications is inconsistent, this was not used as a measure of illness severity.

All charges were converted into costs using hospital-specific cost-to-charge ratios,⁷ adjusted to a US average using cost-of-living adjustment indices for each area,⁸ and converted into year 2000 US dollars using the medical care component of the Consumer Price Index for all urban consumers.⁹

In-patient costs

Per patient hospitalization charges were categorized as medication, procedures, laboratory tests, personnel, room and board, supplies, and miscellaneous. We estimated in-patient physician costs from the number of in-patient days, assuming that a physician saw each patient once a day, multiplied by the Medicare-allowable charge for initial hospital care.¹⁰

Out-patient costs

We estimated out-patient physician costs based on the number of out-patient days (in-patient days subtracted from total TB treatment length). We assumed that after each TB hospitalization, out-patients visited the clinic to see a physician initially every 2 weeks for the first 6 weeks and then once a month until treatment termination. We used the Medicare-allowable physician charge for an established patient.¹⁰ To estimate nurse or outreach worker costs of visiting the patients to provide daily DOT for the remainder of treatment, we used the Medicare-allowable charge for a home visit for an established patient.¹⁰

Because we did not have records of the final drug regimens taken by patients, since they might have varied from those taken during hospitalization, out-patient medication costs were estimated from the number of out-patient days and the cost of the most probable regimen for each patient according to his/her drug resistance pattern. We used doses for an average-sized person (50–74 kg) and applied estimates of Public Health Service prices for the medications.

Indirect costs

We estimated productivity losses for survivors based on the number of out-patient clinic visits and days and severity of illness. We assumed that patients lost a half day of productive work time for each out-patient clinic visit to see a physician, and 25% of their productivity if mildly ill and 50% if moderately ill for each non-clinic-visit out-patient day. We used earnings for non-supervisory production workers, adjusted for fringe benefits and leave,¹¹ converted wages into year 2000 US dollars,⁹ and then converted them to their present value (from the day therapy started) using a 3% discount rate.¹² This resulted in an average of approximately $112.50 per day. Productivity losses were calculated for all patients, including those aged over 65, to include the value of their lost time, whether spent at work in the formal or informal sector, in household production, or at leisure.

To determine the value of life lost for deceased patients, we used human capital productivity loss estimates based on the present value of combined expected lifetime income and household production services.¹³ In the absence of cause of death data, all the value of lives lost in the non-HIV-infected patients and approximately 44% of the value in HIV-infected patients were attributed to MDR-TB. This estimate of 44% for HIV-infected patients is derived from a survival study of a 1997 cohort of HIV-infected TB patients who were sputum smear-positive for acid-fast bacilli at the time of death.¹⁴

RESULTS

The demographics of this MDR-TB group were similar to those of all MDR-TB patients at the time of the study. However, these patients were more likely to be over 65 years of age and to have died during treatment (Table 1). The 13 patients were aged 25–70 years; two were females; and six were not US-born. Seven completed TB treatment, and the remainder died during treatment (Table 2). The aggregate length of hospital stay varied between 5 and 90 days; one was hospitalized three times, and four were hospitalized twice. The length of therapy varied between 140 and 1084 days.

Table 1.

Comparison of study cohort with US surveillance data for 1993–1998

Characteristic	Study cohort %	US surveillance %
Age, years
0–14	0	2
15–24	0	8
25–44	54	53
45–64	23	26
65+	23	10
Foreign-born	46	41
HIV-infected	38	33
Pulmonary TB	92	90
Died during treatment	46	34
Prior TB	15	18

HIV = human immunodeficiency virus; TB = tuberculosis.

Table 2.

Patient characteristics

Severity	Age (years)	Sex	Foreign born	Race/ethnicity	Smear positive	Pulmonary	Cavitary	Treatment	HIV status	Therapy length days	Hospital days	Out-patient days
Mild	67	M	N	White	N	Y	N	Completed	Negative	298	11	287
	70	F	N	Black	Y	Y	N	Completed	Negative	237	20	217
	67	M	Y	Asian	Y	Y	N	Completed	Unknown	295	5	290
Moderate	60	F	Y	Asian	Y	Y	Y	Completed	Unknown	715	85	630
	48	M	N	White	Y	Y	N	Completed	Negative	893	42	851
	25	M	N	Black	Y	Y	Y	Completed	Positive	1084	90	994
	30	M	Y	Hispanic	N	Y	Y	Completed	Unknown	588	8	580
Severe	28	M	N	White	N	N	N	Died	Positive	439	14	425
	35	M	Y	Hispanic	N	Y	N	Died	Positive	270	19	251
	34	M	Y	Hispanic	Y	Y	N	Died	Positive	571	33	538
	37	M	N	Hispanic	Y	Y	Y	Died	Negative	302	14	288
	44	M	Y	Hispanic	N	Y	N	Died	Positive	175	7	168
	57	M	N	Hispanic	N	Y	N	Died	Negative	140	11	129

M = male; N = no; Y= yes; F = female.

In-patient costs (including physician costs) ranged from $5278 to $73 572 (average $25 853; median $16 920) per person, and costs per person-day of hospitalization from $517 to $2053 (Table 3). Room costs comprised the highest proportion of in-patient costs for nine patients (range 25–60%), while medication costs contributed the most for three patients (at 38%, 40%, and 46%). (Figure). Out-patient costs (for medication, physicians, and DOT) per person ranged from $5744 to $41 821 (average $19 028; median $13 210), averaging approximately $44 a day. Direct medical costs (in-patient and out-patient costs) ranged from $12 495 to $115 393 (average $44 881; median $34 103) per person.

Table 3.

Costs of MDR-TB by patient severity of illness

Severity	In-patient	Out-patient	Direct	Indirect	Total
Mild	10 067	13 107	23 174	12 364	35 538
	16 920	9 618	26 538	9 208	35 746
	5 691	13 210	18 901	9 316	28 217
Moderate	43 956	30 113	74 069	46 149	120 218
	68 169	36 060	104 229	52 945	157 174
	73 572	41 821	115 393	66 099	181 492
	9 646	24 457	34 103	34 669	68 772
Severe	28 737	19 477	48 214	621 935	670 149
	24 781	10 647	35 428	552 814	588 242
	33 304	22 738	56 042	604 244	660 286
	8 516	13 155	21 671	1 256 395	1 278 066
	5 278	7 217	12 495	496 995	509 490
	7 449	5 744	13 193	585 903	599 096

MDR-TB = multidrug-resistant tuberculosis.

Figure.

Composition of in-patient costs.

Indirect costs or productivity losses for those who survived ranged from $9208 to $66 099 per person (average $32 964; median $34 669). Indirect costs or the value of life lost for those who died ranged from $496 995 to $1 256 395 per person (average $686 381; median $595 074).

Direct costs were higher for moderate than for severe MDR-TB patients (median $89149 vs. $28550) due to survival and thus longer lengths of therapy (median 804 vs. 286 days) and longer periods of hospitalization (median 64 vs. 14 days). Indirect costs were higher for moderate than for mild MDR-TB patients (median $49 547 vs. $9316) also due to longer lengths of therapy (804 vs. 295 days).

Total costs (direct and indirect) per person ranged from $28217 to $181492 for those who survived (average $89594; median $68772), and from $509490 to $1 278 066 for those who died (average $717 555; median $629 691). Total costs were lowest for mild (median $35 538) and highest for severe (median $629 691) MDR-TB patients. For those who died, indirect costs or the value of life lost comprised 92–98% of their total costs. For those who survived, indirect costs or productivity losses contributed a relatively smaller portion (ranging from 26% to 35% for mild and 34% to 50% for moderate).

Other costs were not included: incentives and enablers, out-of-pocket expenses incurred by patients and their families, costs associated with the transmission of MDR-TB to contacts, and intangible costs such as pain and suffering of patients and their families. Quality of life is likely affected by MDR-TB and was not included in the analysis, other than in reduced productivity estimates. A separate study would be necessary to analyse reductions in quality of life due to MDR-TB.

DISCUSSION

MDR-TB is expensive to patients, society, and the health care system. Because of MDR-TB, patients suffer from a debilitating illness that generally takes a long time to cure and may prove fatal. Society loses the productivity of those affected by the disease, especially of those who die from it. Long hospitalizations, with the potential for secondary transmission of resistant strains, and expensive medications result in high in-patient resource utilization and costs to the health care system. Protracted out-patient treatment can substantially increase the time and resources expended by nurses and outreach workers.

Hospitalization costs for MDR-TB patients were about $9000 higher on average per patient than those reported for all TB patients in the CDC hospitalization study. In this small group of 13 MDR-TB patients, total costs per person averaged $89 594 for those who survived, and $717 555 for those who died. Deaths during MDR-TB treatment were 15% for HIV-negative and 69% for HIV-positive patients during the period 1993–1996,¹⁵ which declined to 8% and 40%, respectively, during the period 1997–1999 (M Moore, CDC, personal communication, 2003). Improvements in HIV care, with highly active antiretroviral treatment (HAART), and in MDR-TB treatment, with increased susceptibility testing and rapid placement on appropriate regimens, likely played crucial roles in these declines. While the number of deaths from MDR-TB has declined since this study was conducted, our estimated costs of death from MDR-TB are expected to be applicable today. Although we estimated MDR-TB costs as comprehensively as possible for the 13 patients, these costs are likely to represent underestimates, as other costs such as out-of-pocket expenses and costs of pain and suffering are not included. However, our MDR-TB cost estimates are valuable in providing a minimum boundary for the cost of interventions needed to prevent MDR-TB development.

The major limitation of this study was its relatively small cohort size (13 MDR-TB patients). In 1995 there were 322 MDR-TB patients diagnosed in the US; therefore, these 13 MDR-TB patients represent 4% of the total for that year. Despite this limitation, this study provides better estimates than currently exist in the literature.

CONCLUSIONS

The societal costs of multidrug-resistant tuberculosis varied mostly because of the length of therapy (including in-patient) and deaths during treatment. With better TB and HIV management, it might be possible to shorten therapy length (including in-patient), avoid deaths, and thus reduce the overall cost. By conducting drug susceptibility tests on specimen cultures, by choosing appropriate treatment regimens, and by managing patients to increase treatment adherence (through individual case management including DOT), physicians can prevent the development of secondary (or acquired) drug resistance and minimize therapy length. Primary MDR-TB disease transmission and subsequent care and deaths can be reduced through use of negative-pressure rooms to isolate MDR-TB patients. On recognition of MDR-TB, and especially in HIV-infected patients, patient care management should be undertaken in close consultation with MDR-TB experts.¹