OBRA 87

OBRA 87 has been characterized as a "watershed"; it provided a definition of quality in long-term care that focused measurement of quality on resident outcomes and resident rights, and it recognized that without appropriate attitude and motivation, quality of care cannot be provided (Wilging, 1992a, p. 22). OBRA 87 specified that a nursing facility "must provide services and activities to attain or maintain the highest practicable physical, mental, and psychosocial well-being of each resident in accordance with a written plan of care …" (sec. 1919(b)(2)). This legislation was a "landmark" statute in at least three respects.

First, for the first time, nursing homes receiving federal funds were required to ensure a high quality of life in addition to providing a high quality of care.

Second, the requirement that well-being be maximized implied that improvements in health and functional status be achieved, when possible, thereby shifting the focus from provision of custodial to provision of rehabilitative care. To achieve the objective of the highest level of well-being, nursing homes were required to develop individual care plans and a resident assessment process. Under the provisions of OBRA 87 and federal guidelines, nursing homes participating in Medicare and Medicaid programs must use the resident assessment instrument (RAI) to assess residents on admission, annually thereafter, and on any significant change in the resident's status. The RAI consists of the minimum data set (MDS) for resident care assessment and care screening (which is described more fully below) and resident assessment protocols (RAP). The purpose of these assessments is to identify a resident's strengths, preferences, and needs in key areas of functioning and to guide the development of the resident's care plan (Phillips et al., 1994).

Third, OBRA 87 recognized that implementation of its provisions would require additional resources and training. Therefore, it encouraged state Medicaid programs to adjust their rates to reflect the new OBRA standards. This is arguably the first time that Congress explicitly recognized that high-quality care and quality assurance efforts come at a price.

HCFA Regulations

HCFA issued the enabling regulations in October 1990. These regulations mandated a number of changes. First, the regulations eliminated the hierarchy of conditions, standards, and elements that had been in prior regulations to that point.

Second, the 1990 regulations mandated comprehensive assessments of all nursing home residents, using the MDS forms (Morris et al., 1990). Nursing facilities are required to complete the MDS forms for each resident within 14 days of admission, when there are major changes in health status, and at least annually. Facilities also are required to use the assessment in the care planning process. The federal survey procedures (conducted by state agencies) check the accuracy and appropriateness of the assessment and care planning process for a sample of residents.

Third, more specific requirements for nursing, medical, and psychosocial services were designed to attain and maintain the highest practicable mental and physical functional status (Zimmerman, 1990). These requirements were specified in the new regulations, and a detailed set of HCFA interpretive guidelines was developed for use by state surveyors in 1990. The state surveys were redesigned to be more outcomes oriented than had previously been the case. Such outcome measures include residents' behavior, their functional and mental status, and certain physical conditions (such as incontinence, immobility, and decubitus ulcers). For example, the regulations established criteria for and prohibited the use of physical restraints and antipsychotic drugs without a specific indication of need, and they require periodic review and dose reduction unless clinically contraindicated. In addition, regulations detailing and protecting residents' rights were added.

Nursing Home Resident Assessment Tools

An important recent advance stemming from OBRA 87 was the development of the Nursing Home Resident Assessment System by Hawes, Phillips, and their colleagues. This system designed the nursing home minimum data set mentioned above for resident assessment and developed detailed protocols for resident assessment of specific problem areas to guide the care planning process (Morris et al., 1990). The purpose is to assess the functional, cognitive, and affective levels of residents. The MDS items were field-tested in 1990; the final version included 15 domains: cognitive patterns, communication/hearing patterns, vision patterns, physical functioning and structural problems, continence, psychosocial well-being, mood and behavior patterns, activity pursuit patterns, disease diagnoses, health conditions, oral/nutritional status, oral/dental status, skin condition, medication use, and special treatments and procedures (Morris et al., 1990).

Now under development are quality indicators (QI), which use the MDS as a part of the National Nursing Home Case-Mix and Quality Demonstration (NHCMQ) study funded by HCFA. Among them are QIs for accidents, behavioral/emotional problems, cognitive problems, incontinence, psychotropic drugs, decubitus ulcers, physical restraints, weight problems, and infections. The QIs for individual residents and for facilities are compared to national norms, by taking into account predisposing factors and case-mix factors related to each QI. QIs that may indicate poor quality of care are identified and given to state surveyors to examine in the certification survey process (Zimmerman et al., 1995). Using QI data, state surveyors are expected to determine whether or not the identified QIs are the result of or are related to poor care processes.

HCFA has proposed issuing regulations to require all nursing facilities to store and transmit RAI information electronically. Although, as of late 1995, the final rule had not been published, most providers are proceeding on their own. Nearly 62 percent of nursing facilities have begun computerizing resident assessments (AHCA, 1995). When fully computerized, QIs may be a valuable tool for monitoring the quality of nursing home care. Certainly, the QIs will augment the current nursing home survey process.

Survey and Certification

In November 1994, HCFA (1994a) released its final regulations for the survey, certification and enforcement of SNFs and NFs (42 CFR Parts 401–498). The provisions shaped the process of surveying and certifying facilities and specified procedures for enforcement. HCFA is also undertaking efforts to train state surveyors in using the new survey, certification, and enforcement procedures.

Several alternative remedies may be imposed on facilities that do not comply with federal requirements, instead of or in addition to termination. These include civil money penalties of up to $10,000, denial of payment for new admissions, state monitoring, temporary management, immediate termination, and other approaches. The extent and type of enforcement actions depend on the scope of problems (whether deficiencies are isolated, constitute a pattern, or are widespread) and the severity of violations (whether there is harm or jeopardy to residents).

Federal Standards

To the extent that new federal requirements were binding on facilities, some improvement in staffing and the consequent quality might be anticipated. OBRA 87 required nursing facilities to have licensed nurses on duty 24 hours a day; a registered nurse (RN) on duty at least 8 hours a day, 7 days a week; and an RN director of nursing. The statute permits the director of nursing and the RN on staff for 8 hours to be the same individual. In addition, OBRA 87 specifies that facilities have sufficient staff to accomplish the care objectives described above (sec. 1919(4)(C)(i)).

As stated in chapter 4, Staffing ratios have increased slowly in recent years (HCIA and Arthur Andersen, 1994). This slight improvement can be attributed in part to the requirements of OBRA 87 and in part to the staffing need to care for residents who require specialized services (such as subacute care and services for residents with Alzheimer's disease).

Any effect of the nurse coverage requirements may have been mitigated in some circumstances by the waiver provisions of OBRA 87. Because of the nursing shortage at that time, Section 1919(b(4)(C)(ii) of the Social Security Act authorizes any state to waive the requirements for 24-hour licensed nursing service as well as the 8-hour-a-day RN presence if certain criteria are met. The facility is required to demonstrate that, despite diligent efforts, it has been unable to recruit appropriate personnel. The state must also determine that a waiver will not endanger the health and safety of individuals residing in the facility. Waivers must be obtained by facilities annually.

Quite a large number of facilities have obtained waivers. The Secretary of DHHS reported to Congress on initial waivers granted through January 1993. In 1993, governments in 13 states granted waivers to 518 of the 5,302 facilities certified for Medicaid only (HCFA, 1994b). Of these facilities, 66 were waived from the requirement to provide 24-hour licensed nurse services. More prevalent are waivers for the 8-hour RN requirement. As of January 31, 1993, the states had granted such waivers to 490 facilities. Some progress has been made in the intervening period. For example, by May 1995 the number of waivers granted dropped to 157. All of them were for the requirement for RN presence.³

The committee is encouraged by the progress made by states in reducing the number of waivers, especially those related to 24-hour licensed nurse presence. Since increasingly sicker and older persons requiring skilled nursing care are being discharged from the hospitals to nursing facilities, more needs to be done by the facilities to ensure an RN presence to provide care for the residents. With the adequacy of the RN supply at the present time and reported layoffs in hospitals, nursing facilities should have less difficulty in meeting this statutory requirement without waivers. The committee strongly endorses the intent of OBRA 87 and supports efforts made by facilities and states to improve professional nurse staffing in nursing homes consistent with the intent of the statute. The committee hopes that the states continue the progress made in reducing the waivers granted and that they work diligently toward eliminating them in the foreseeable future.

State Standards

In addition to federal standards, some states have established their own standards, although these may vary widely across states. Mohler (1993) surveyed states regarding their staffing requirements for nursing facilities. She found that the majority of states had specific minimum staffing standards in addition to the federal standards for nursing facilities. Some states specified standards for RNs and others for nurse assistants; still others had standards for both. For example, Minnesota requires a minimum of 2 hours of nursing care per resident day for all licensed nursing facilities, although these hours are not required to be distributed evenly across the evening or night shifts (Chapin and Silloway, 1992). States can also impose state penalties on facilities that have substandard staffing based on state regulations. These data suggest that many states have concluded that they need staffing standards for nursing facilities in addition to those of the federal government.

Empirical Evidence on Staffing

The committee reviewed the available data and research literature that provide empirical evidence of changes in staffing in nursing homes that may have occurred since the enactment of OBRA 87. Comparing data from the 1991 National Health Provider Inventory with corresponding data from the 1985 National Nursing Home Survey (NCHS, 1987; HRSA, 1994), the committee determined that overall staffing ratios for RN time and total staff time per resident day had increased very slightly over the 18 minutes of RN time and 2 hours of total staff time per resident day reported in 1985 (NCHS, 1988). Moreover, most estimates assume that staff are evenly distributed over 24 hours, which is generally not the situation. Most facilities have fewer staff (both licensed and total staff) on evening and night shifts and on holidays and weekends than during daylight or ''regular" hours of operation.

Kane reported that Friedlob's 1993 doctoral dissertation analyzed the results of a large sample of nursing homes and their residents in 6 states (R.A. Kane, 1995). About 39 percent of the sample received no care from an RN during the 24-hour study period; the average RN time per resident was 7.9 minutes; the average licensed practical nurse (LPN) time, 15.5 minutes; the average nurse assistant (NA) time, 76.9 minutes.

Zinn (1993a) found wide variations in nursing home staffing patterns in 10 standard metropolitan statistical areas (SMSA) studies, even after controlling for case-mix.

Current staffing data collected by state surveys are available from the federal OSCAR. As shown in Table 6.4, between 1991 and 1993, RN hours or total nursing hours per resident day in facilities certified as Medicaid-only did not change. RNs continue to report 0.3 hour (or 18 minutes) per resident day, and LPNs 0.6 hour. In Medicare-only and Medicare and Medicaid dual-certified facilities, staffing levels for all categories of nursing increased appreciably over the 3-year period. The committee notes, however, that the vast majority of facilities were not Medicare certified during this period, and those not certified by Medicare had a much lower staffing base. Thus, staffing levels improved somewhat for dual-certified facilities but remained essentially unchanged for Medicaid-only facilities. The differences reflect, at least to some extent, differences in acuity and reimbursement levels.

TABLE 6.4

Nurse Staffing Levels for All Certified Nursing Facilities from the Federal On-Line Survey Certification and Reporting System, 1991–1993.

Ratios for the country as a whole obscure substantial variation among states and among facilities within states. OSCAR data show that in 1993, 96 facilities in 20 states reported having no RN staff. It is not known if this was the result of reporting errors or represented a real absence of RN staff. Reporting problems with the current OSCAR data on staffing have been identified. Actual staffing levels in nursing facilities may be lower or higher than the levels reported because of over- or underreporting and other errors. Staffing data are reported by the facilities to state surveyors and are not always reviewed by the surveyors.

Public Hearings and Site Visits

The committee conducted public hearings and site visits to learn about these issues directly from the persons and organizations affected. Many complaints were voiced at the public hearings by resident advocates, licensed nurses, and nurse assistants about inadequate quality of care and staffing shortages at all levels. Representatives of the nursing home industry, however, described improvements in the quality of nursing care indicating that, in their view, staffing was on average adequate. They also described the financial and other constraints under which facilities operate.

The committee cannot generalize from such testimony. Rather it notes the lack of current, nationally representative, and valid data on the characteristics of nursing facilities, their residents, and the staffing in these facilities, as well as on processes and outcomes of care in nursing facilities. The last comprehensive National Nursing Home Survey was conducted in 1985. The data from this survey have been used extensively by researchers, policymakers, industry, and resident groups. The survey provided detailed information on nursing facilities, staffing, discharges, and patient characteristics. Primarily because of budget constraints, this survey was not conducted again until a decade later in July 1995. The committee regrets, however, that because of budgetary problems, the survey instrument has been pared down. The committee understands that critical detailed information on staffing and discharges such as that collected in 1985 will not be collected. Only total numbers of full-time equivalent RNs, LPNs, and NAs will be collected. Current, comprehensive data are needed for research, policy formulation, management of the nursing home industry, and consumer information.

Information derived from administrative databases such as the OSCAR files and similar sources should be made available in a format that is reliable and understandable to the public. Every effort should be made to ensure the quality and accuracy of these administrative files, to adequately train the persons surveying and reporting the information, and to ensure that the information is used appropriately. These issues surrounding the availability of reliable, accurate, valid, and timely data need attention in order to ensure that policies are made and evaluated on a solid factual and analytical basis.

During its site visits, the committee repeatedly heard complaints from nursing home staff about the paperwork imposed by OBRA 87 and the states. In particular, they indicated that a substantial part of the RN's time was spent completing paperwork associated with the MDS, so that staff had insufficient time for direct patient care. The link between the completion of the MDS and the provision of high quality of care was not always apparent in every facility. Data from such assessments are essential to the development of outcome-oriented measures of quality and the implementation of a resident-focused federal certification process. Some viewed the MDS as another intrusion imposed by external sources on their time rather than as a clinical tool for improving the quality of resident care and the management of patient care. Also, committee members noted that at some sites, when the staff complained about the burden of MDS, they did not distinguish between the MDS and other state and local forms that were merged with it. The general belief among some of those interviewed was that they provide quality care and that quality can be better improved with more time devoted to patient care instead of paperwork. Clearly, social engineering is needed to create a different climate and professional nurses should be central to this endeavor.

Although the committee is sympathetic about the time-consuming nature of the forms, it does not conclude that the existing implementation is unnecessarily burdensome. The committee is concerned, however, about the apparent lack of consensus among nursing home staff in recognizing that such clinical tools are needed at the grass-roots level. Some have also questioned the quality of the information entered on the MDS because of an emphasis on meeting regulatory survey requirements. The MDS is supposed to reflect accurately what happens so that appropriate interventions are undertaken in a timely manner. It is meant to be a link among assessment, planning, intervention, and evaluation, but it is seen by many staff in the nursing facilities as a legal requirement that they have to complete. In some instances, the nurse who is required to complete all the care plans based on the MDS may not ever see the patients in question. Some nursing facilities hire consultant nurses for the purpose of completing the MDS and other documentation (Schnelle, 1994).⁴ The committee believes that use of consultant nurses for this purpose only is undesirable, and it strongly discourages facilities from continuing the practice.

The committee strongly endorses the concept of an individualized care plan for each resident. Such care planning requires the use of tools like the MDS. A uniform, comprehensive resident assessment system is essential to the development of an individualized care plan for each resident that focuses on improving, maintaining, or minimizing decline in the resident's functional status and quality of life.

The committee endorses HCFA's current efforts at improving the instrument and requiring all facilities to computerize MDS data and provide them to state and federal agencies, thus providing a mechanism for a national database. The committee further encourages the states to adopt use of the MDS without unduly adding additional elements to it.

Staffing and Resident Characteristics

There is universal agreement in the research literature on the strong relationship among resident characteristics, nursing staff time requirements, and nursing costs in nursing facilities. Several studies have examined these relationships and attempted to quantify them (Weissert et al., 1983; Arling et al., 1987). Case-mix is an important factor in examining the relationships between research needs and nursing staff required to meet those needs. Resident characteristics were studied in terms of staffing resources in facilities judged to offer high quality of care in the development of the Resources Utilization Groups (RUGS) (Fries and Cooney, 1985). Additional studies were used to create an updated RUGS II (Schneider et al., 1988). Substantial work has been conducted as part of the Multistate Nursing Home Case-mix and Quality Demonstration Project sponsored by HCFA beginning in 1989 (Fries et al., 1994). The basic principle of the case-mix project is that resources should be allocated based on resident need and that sicker and more debilitated residents need more services both in terms of amount of staff time as well as level of expertise. As a result of research into case-mix, the third iteration of a resource utilization grouping (RUGS III) has been developed (Fries et al., 1994). RUGS III was developed with 44 resident groups, which were defined to explain 56 percent of the resource utilization variance (Fries et al., 1994). Based on the data from the case-mix project, it is evident that the need for nursing time and nursing skills varies significantly depending on the resource classifications. These data show that facilities need to adjust their staffing levels to take into account the condition of the residents. Fries and colleagues' (1994) research has documented that much of staffing is driven by the type of residents.

The use of staffing standards or ratios as a structural indicator of the quality of nursing homes presupposes that higher ratios lead to improved care processes and outcomes. Although it is not clear that any fixed staffing-to-resident ratios can be established across all facilities and all types of residents, facilities should use "acuity indexes" or other case-mix methods for adjusting their staffing levels to ensure sufficient staff to provide for the basic needs of residents.

Staffing and Quality of Care

Considerable attention has been devoted by researchers and policymakers to the issues of how many staff and what type of staff are needed to meet the needs and expectations of nursing home residents as measured by care processes and outcomes. Several studies have been conducted by experts in the field to examine these relationships. They show a strong relationship between both general level of nursing staff and RN staff (or professional nursing staff, in particular) and resident outcomes.

Not surprisingly, higher nursing staff levels (nursing staff hours per resident day) in nursing facilities have been associated with higher quality of care as measured in terms of care processes and various outcomes. Type of nursing staff may be more important than the availability nursing staff hours per se. Nursing care is a major service provided by nursing homes. Some experts believe that nursing facilities that rely predominantly on unskilled nursing staff with minimal presence of licensed nursing staff jeopardize the quality of nursing home care.

The studies reported below examine the relationships between the nursing staff and other nursing home characteristics as independent variables and quality of care as measured in terms of care processes and resident outcomes as the dependent variables. Nursing staff as an independent variable is examined in various ways in these studies. Some researchers have used total numbers of nursing staff or aggregate nursing staff levels, others have examined the number or level of RNs (referred to at times as professional nursing staff), still others have looked at staffing as the ratio of RNs to LPNs or RN to total nursing staff (sometimes referred to as skill mix). Table 6.1 provides an illustrative list of structure, process, and outcome measures used to measure quality of care in nursing homes.

In a landmark study by Linn and colleagues (1977), conducted over a 9-year period, 1,000 men transferred from a general medical hospital into 40 community nursing homes were studied. Information was gathered about the men at the time of discharge from the hospital, 1 week after transfer, and 6 months later. The purpose of the study was to determine the relationship of nursing home characteristics to differential outcomes of patients placed in several homes. Patients' outcomes measured were mortality, functional status, and discharge from the nursing home. The independent variables were nursing home characteristics—an array of structural variables such as facility size, ratios of RN hours per patient, LPN hours per patient, NA hours per patient, total staff hours per patient, professional staff hours per patient, costs per month, medical records, meals, patient appearance, services, policies, physical plant, and safety. Outcomes were related to the nursing home characteristics by multivariate analysis of variance, controlling for expected outcomes—age, and diagnoses of cancer and chronic brain disease. The authors found that a higher ratio of RN hours per patient was consistently and significantly associated with all three outcome measures: patient survival, improved functional status, and discharge from the nursing home. Hours per patient day of the other care givers or the total staff to patient ratio were not related to outcomes. Meal service was related to mortality and functional status; medical records, higher professional staff to patient ratios, and services were related to discharge from the nursing home. Over half of the remaining nursing home variables were never associated significantly with any of the patient outcomes. The authors conclude that increasing RN hours and giving more attention to meal service will have a positive impact on patient outcomes.

Nyman (1988b) examined if quality requires more financial and physical resources using the 1983 Iowa Outcome Oriented Survey. Eight nonhealth quality variables were used, including plant maintenance, room maintenance, room furnishings, care plan, diet plan, Medicare plan, resident care, and quality of life. The independent variables included licensed nurse hours, NA hours, administrative hours, for-profit status, number of beds and other variables. Nurse hours per patient were positively related to seven of the eight quality variables, and significantly related to three. NA hours per patient or any other labor input were not positively or significantly to the quality variables. These findings suggest that simply requiring more staffing in general may not be sufficient to insure quality. None of the quality measures constructed from these data were significantly related to higher average costs.

Using data from reports of 455 Medicare-certified skilled nursing facilities, and controlling for case-mix, Munroe (1990) examined the relationship between facility quality and nursing personnel. The purpose of the study was to determine the extent to which RN staffing patterns influenced nursing home quality. He found a positive significant relationship between nursing home quality as measured by numbers of health-related deficiencies received by the facilities and a higher ratio of RNs to LPNs hours per resident day after controlling for several variables including case-mix. The author concluded that the configuration (or staff mix) of nursing personnel may be more important than total nursing hours.

Based on analysis of 2,500 nursing home residents in 80 nursing facilities in Rhode Island, Spector and Takada (1991) used multivariate models to estimate what structure and process variables are associated with resident outcomes after controlling for resident characteristics. Outcomes measured over a 6-month period included mortality, functional decline, and functional improvement. They found that higher staff levels and lower RN turnover were related to functional improvement and found that low staffing in homes with very dependent residents was associated with reduced likelihood of improvement. High urinary catheter use, low rates of skin care, and low resident participation rates in organized activities were all associated with poor resident outcomes. Overall, few process or structure variables were significantly related to mortality.

Cohen and Spector (in press) found that staffing ratios per residents have a significant impact on resident outcomes. Their analysis of a nationally representative sample of nursing homes and nursing home residents highlights the complex nature of the relationship between RN and LPN staffing and resident outcomes, and shows that staff mix is more important than the overall numbers of staff. Using regression analysis, the authors estimated the relationship between "staff intensity" and resident outcomes to determine if in fact more intensive staffing results in better outcomes. Staff intensity in this study is measured as the number of full-time-equivalent (FTE) staff per 100 residents adjusted for case-mix. Three staffing equations were estimated—RNs, LPNs, and total nursing staff. Three resident outcome equations were estimated. The dependent variables were: mortality within a year, having a bedsore, and functioning defined as ADL status at the end of the year, controlling for health status. Staff intensity effects were found for both mortality and ADL outcomes, but not for bedsores. A higher RN intensity was associated with a lower rate of mortality. The authors calculate that addition of half of an FTE RN (about 10 percent increase in RN on average staffing) would save about 3,000 lives annually. A higher intensity of LPN staffing improved functional status as measured by ADLs, although this impact was relatively small. Having more NAs had no impact on resident outcomes, at least the ones measured in this study. The authors suggest that less variation in NA numbers across nursing homes than in RNs and LPNs may explain to some extent the failure to observe a NA staffing effect on quality.

The finding that RNs affect mortality and LPNs affect functional outcomes is consistent with other research findings. RNs are more trained than LPNs. RNs provide leadership in terms of the organization of the entire nursing care in the facility. They are the ones trained to identify early a potential life threatening situation, and recognize regression of condition in a resident. In a medical emergency an RN deals with the problem until a physician is consulted. LPNs on the other hand are more involved in the day-to-day nursing care than RNs, and they are better trained than NAs in nursing procedures and routines. These results suggest that RNs and LPNs are not substitutes, but have different values for the outcomes of residents.

The authors conclude that since professional nurse staffing intensity has a direct impact on quality, as measured by resident outcomes, quality can be improved by directly influencing the mix of staffing. Efforts to improve quality should be focused more on increasing the intensity of professional nursing staff than only on nonprofessional nursing staff. The authors caution not to interpret the data to mean that the level of NA staffing is unimportant; they do suggest, however, that efforts to improve quality should be focused more on increasing the intensity of the professional than only on the nonprofessional staff.

Zinn (1993b) conducted a study using data on approximately 14,000 nursing facilities from the 1987 Medicare and Medicaid Automated Certification Survey (MMACS). Using a weighted two-stage least square regression model and controlling for case-mix, she found that a lower ratio of RN hours to other nursing staff hours was associated with greater use of urinary catheters, physical restraints, and tube feedings, and with less toileting of residents.

Braun (1991), in a retrospective cohort study of 390 veterans discharged to 11 nursing homes and followed for 6 months investigated the relationship between nursing home quality and patient outcomes of mortality, rehospitalization, and discharge, controlling for severity of illness and case-mix differences. The Multiphasic Environmental Assessment Procedure (MEAP) was used as for quality assessment. Additional quality indices were used as independent variables such as process of nursing care, RN hours, equipment, and Medicaid status. The results of the analysis show that the quality-of-care variable "RN hours" was significantly and inversely related to mortality, while the quality-of-care variable "use of nursing process" was significantly related to probability of discharge.

Using data originally collected by the Missouri Division of Aging during its routine inspection of nursing homes, Cherry (1991) studied 134 Medicaid and Medicare certified nursing homes in Missouri to assess the role of the ombudsman in nursing home quality as measured by the various health related outcomes. In addition to the ombudsmen program, the independent variables included RN hours per resident day, LPN hours per resident day, NA hours per resident day, case-mix, profit, percent of Medicaid residents, and facility size. The only variable contributing significantly to improved quality of care was RN hours per resident day. Increased RN hours were positively associated with improved quality of care measured through a composite of outcome indicators including number of residents developing decubitus ulcers per immobile residents, number of residents catheterized per incontinent residents, number of urinary tract infections per incontinent residents, and rate of antibiotics use per resident. In contrast, hiring more LPNs and in particular more NAs did not necessarily yield similar increases in quality. Adding more NAs also increases the need for supervision and is not in itself a solution to alleviating poor care. The presence of an ombudsman program was found to be significantly associated with quality for skilled nursing homes where there was ample RN staffing.

Gustafson and colleagues (1990) found a significant correlation between nursing staff levels and six measures of quality incorporated into the Quality Assessment Index constructed by them for measuring nursing home quality. A study of nursing homes in Maryland found that higher total staff levels are related to fewer nursing deficiencies (Johnson-Pawlson, 1993).

An anthropological study conducted by Kayser-Jones and colleagues (1989) analyzes the clinical and sociostructural-structural factors contributing to the hospitalization of nursing home residents. Data were collected by participant observation and event analysis. Qualitative analysis of the data found that "insufficient" and "inadequately trained" nursing staff were contributing factors to the deterioration and eventual hospitalization of nursing home residents. A large proportion of the patients could have been treated in the nursing home if the nursing staff had been available and able to administer IV therapy and to monitor the response to treatment and effectively communicate with the physician. Most of the nursing care was provided by a small number of LPNs, but mostly by NAs. The shortage of nursing staff also contributed to nutritional problems. Staffing is an important variable that influences eating behavior. Inadequate staffing results in (1) the feeding of residents in a hurried manner that does not preserve their dignity (e.g., giving residents a large amount of food with each bite, feeding several residents at once, mixing food); and (2) inadequate nutritional intake, resulting in resident weight loss and necessitating the use of liquid supplements and sometimes tube feedings with all its attendant risks. These findings are supported by the research of Blaum and colleagues (1995), which showed poor resident nutrition associated with being dependent on staff for feeding.

Several clinical intervention studies support the need for professional nurse presence to provide the leadership and direction for assisting staff in order to assure that the success of interventions efforts as measured by reductions in physical restraints, presence of bed sores, or falls, continues beyond completion of the study period (see, for example, Evans and Strumpf [1994] and Mezey and Lynaugh [1989] discussed in the next section).

Some researchers (Schnelle, 1990; Schnelle et al., 1990; Hawkins et al., 1992) tested staff management procedures and policies (such as oral and written feedback, professional staff and gerontology specialists working along with the NAs and guiding and training them, and the continuing presence of the supervisory professional nurse even after the intervention period) for assuring that NAs are appropriately trained and motivated to carry out the intervention protocols for decreasing incontinence. Schnelle (1994) reported that during the period of the studies the investigators gave on-site, on the floor training, and walked the floors with the NAs to train staff and to ensure adherence to protocol. These interventions showed significant improvements in resident dryness. However, the interventions only had a positive effect until the controlled conditions were withdrawn; then staff reverted to previous practices. In a multisite study, the staff management system and incontinence protocol were continued in only one of the sites after the controlled conditions ceased. At this site, the director of nursing was particularly interested and involved, and frequently got out on the floor and actually put her hands on the patients to see how often they were wet. Schnelle concluded that it was the knowledge, hands-on interest, and leadership of the RN that resulted in the sustained use of the intervention protocols.

Given the level of NAs' direct care responsibilities and the minimal training for resident care required for them, professional nurse oversight and availability to work closely through constructive supervision of NAs and other licensed nurses is critical (see Johnson et al., Part II of this report). Johnson and colleagues' analysis of OSCAR data indicates a continuing increase in case-mix since 1992. Clinical care gets complicated as residents are likely to have multiple chronic conditions, which require understanding each illness and the interacting effects. Residents are sometimes on complicated medication and treatment regimens requiring pharmacological knowledge. Hence, as the case-mix increases, resident care requires a highly qualified nursing staff present at all times, with the nurse needing a broad base of knowledge covering basic nursing, geriatrics, rehabilitation, and psychiatric skills. Today, an LPN may be the only licensed nurse in a facility in the evenings and night time to attend to resident care; an LPN does not meet the qualification requirements stated above. Increasing the number of RNs becomes very important as the case-mix index of a facility increases. Johnson and colleagues (Part II of this report), in their analysis of "high- and low-quality" facilities, found a positive relationship between the amount of RN time and quality of care. Given this relationship, RN participation in care is very important. They concluded that the current federal requirement of one RN for 8 hours a day, 7 days a week is not sufficient to ensure quality of care for residents.

The preponderance of evidence from a number of studies using different types of quality measures has shown a positive relationship between nursing staff levels and quality of nursing home care, indicating a strong need to increase the overall level of nursing staff in nursing homes. There is a strong need to improve the overall level of nursing staff (RNs, LPNs, and NAs) in nursing homes, but prescribing a staffing ratio across all residents and facilities is inadequate. Research literature does not answer the question of whether a particular ratio of total nursing staff to residents is optimal. Moreover, varying circumstances among nursing homes, case-mix differential within and between facilities, and other factors, including those described in the next section, also affect the type and level of staffing needed. The committee, therefore, endorses current HCFA staffing standards but is not inclined to recommend a specific minimum staffing ratio across all types of facilities to meet the needs of all types of patients. Nursing facilities, however, should ensure adequate nursing services to meet the acuity needs of its residents.

Based on the empirical evidence presented above, the committee concludes that a relationship between RN-to-resident staffing and quality of care in nursing facilities has been established. Although the committee did not uncover any research specifically testing 24-hour nursing presence in a controlled experiment, there was sufficient evidence in the literature reviewed that the presence of RN (including geriatric nurse specialists) improved quality of care in nursing homes. Research reviewed above provides abundant evidence that the needs of current residents require highly capable staff, consistent with RN education, and that participation of RNs in direct care giving and providing hands-on guidance and supervision to the NAs and LPNs in caring for the residents is positively associated with quality of care. The committee, therefore, supports the need to increase professional nurses in nursing homes on all shifts. Given the findings on the beneficial effects of continuous RN presence on various dimensions of quality, especially on outcomes, and on the use of RNs versus LPNs on the number of health-related deficiencies in nursing homes, the committee supports an increase in the mix of RNs to other nursing staff. Particularly problematic, however, is the lack of any RN presence in many facilities on the evening and night shifts. In light of the data that indicate the fairly low level of education and high turnover rate among NAs in nursing homes, the knowledge and judgment of an RN is critical to recognize a crisis or a regression of a condition. Early detection and intervention related to particular signs and symptoms often forestall the use of more expensive care resources, most notably hospitalization; this is the work of the professional nurse. Based on the above evidence and the committee members' professional experience and expertise, the committee gives priority to mandating improvement in this area of enhancement.

RECOMMENDATION 6-1: The committee recommends that Congress require by the year 2000 a 24-hour presence of registered nurse coverage in nursing facilities as an enhancement of the current 8-hour requirement specified under OBRA 87. It further recommends that payment levels for Medicare and Medicaid be adjusted to enable such staffing to be achieved.

The committee deliberated hard on this issue, and some committee members are hesitant to impose additional costs on the facilities. The committee recognizes that this recommendation entails additional costs and, therefore, is recommending that Medicare and Medicaid reimbursements be adjusted accordingly, once reliable figures on potential additional costs can be derived. The empirical evidence, however, is convincing, and trends for the future suggest an increasing need for professional nursing presence. The committee further believes that waivers to this requirement could be granted by states only in exceptional circumstances. The committee estimates that the cost of the substitution of the additional 16 hours of RN time (over the current 8 hours) for those hours of LPN time across all nursing facilities in the nation could amount to roughly $338 million. This figure is a rough estimate and may be considered an overestimate since most Medicare-certified facilities and some Medicaid-certified facilities already have RNs on 24-hour duty and because of the expectation, based on the geriatric nurse practitioner studies discussed below, that some offsetting cost-savings would be achieved from prevention of complications, higher levels of function, and fewer hospitalization resulting from early detection of signs and symptoms and timely intervention.⁶ Ultimately, the committee believes that the public's interests and needs for basic quality of care must be considered in addition to cost if we as a society are going to maintain a sense of values and responsibility for the care of the elderly, disabled, and disadvantaged. At the same time the committee recognizes the possible hardships for a few facilities, such as those in remote rural areas, to recruit and retain 24-hour RN coverage. One committee member dissented from the committee's decision to recommend the 24-hour presence of registered nurses in nursing facilities. His separate statement is included in Appendix C of this report.

The committee strongly believes that research should continue to further refine the relationships between staffing and resident outcomes, controlling for the relevant intervening factors including structural and organizational variables. The committee also urges continued efforts towards completing the very important case-mix demonstration project under way and going the next step to link the categories being developed of resource needs and measurement of staff time to meet those needs with the quality-of-care measures.

Geriatric Specialists

The IOM Committee on Nursing Home Regulations encouraged nursing homes to employ specialty-trained gerontological nurses and to encourage gerontological nursing (IOM, 1986b).

Studies of gerontological nurse specialists (GNS) and geriatric nurse practitioners (GNP) in nursing homes have shown that they can improve resident outcomes and contribute to quality by changing the focus from custodial to rehabilitative care (Kane et al., 1976, 1988) and by increasing the ability of facilities to care for more complex and acutely ill patients (Mezey and Scanlon, 1988). Employment of GNPs does not adversely affect nursing facility costs or significantly affect profits. There is also some evidence of cost savings, particularly in medical service use by newly admitted patients. GNPs also reduce the use of hospital services (Buchanan et al., 1990).

Evans and Strumpf (1994), investigating the relative effects of two experimental interventions delivered by GNSs on the use of physical restraints and resident and staff outcomes, found that staff mix and resident personal competence were important factors in the occurrence of disturbing behaviors likely to be managed by restraint. These disturbing behaviors occurred more frequently in situations where the availability of licensed nurses was low and resident frailty high. Further, although staff increased their assessment and intervention skills, the investigators noted the need for a ''consistent professional presence" of a clinician with geriatric expertise to maintain minimal restraint use in the facility. Findings suggest that quality outcomes (e.g., restraint reduction) do not necessarily require more staff per se, but do require staff who have the requisite knowledge base, gerontological expertise and education, as well as resident-centered assessment, monitoring, care planning, evaluation, and support in their efforts to provide quality individualized care (Evans and Strumpf, 1994; Strumpf, 1994).

A number of demonstrations have provided convincing evidence that GNPs and GNSs are effective in nursing homes. HCFA supported the evaluation of two demonstration projects, the Robert Wood Johnson Foundation Teaching Nursing Home (TNH) Program and the Nursing Home Connection (Massachusetts 1115: Case Managed Medical Care for Nursing Home Patients), while the Kellogg Foundation supported the Mountain States program to place GNPs in nursing homes. These evaluations confirmed that nurses with advanced preparation in care of the elderly decrease unnecessary hospitalization and use of emergency rooms, improve admission and ongoing patient assessments, provide better illness prevention and case finding, decrease incontinence, lower the use of psychotropic drugs and physical restraints, and generally improve the overall management of chronic and acute health problems. These improvements in care occurred without incurring additional costs and in some instances at a reduced cost.

The experience of the TNH program provides further evidence of the need for professional gerontological nurses in nursing homes (Mezey and Lynaugh, 1989, 1991; Mezey, 1994). In comparing the TNH to matched nursing homes in the same state, with the only difference being the presence of an advanced practice nurse in the former, the residents in TNHs had significantly fewer hospitalizations than those in the comparison homes and fewer emergency room visits. There also were notable improvements in a variety of quality indicators such as the management of urinary incontinence; decreased use of psychotropic medications, including long-acting benzodiazepines and other such medications related to poor quality outcomes in nursing homes; and less use of restraints.

Based on the above review of research, the committee concludes that there is sufficient evidence that presence of geriatric nurse specialists/practitioners enhances quality of care in nursing homes. Moreover, research has shown that cost savings in the long run accrue, particularly due to reduced rehospitalizations and visits to hospital emergency rooms.

RECOMMENDATION 6-2: The committee recommends that nursing facilities use geriatric nurse specialists and geriatric nurse practitioners in both leadership and direct care positions.

Nurse Assistants

Nurse assistants constitute 70 to 90 percent of nursing staff in nursing facilities (IOM, 1986b; NCHS, 1987; Maraldo, 1991). They provide most of the direct care and spend the most time with the residents. More than 90 percent of the NAs are women (BLS, 1995c; Crown et al., 1995). Three-fourths of the NAs in nursing homes have not completed high school. They often come from low-income families, earning close to minimum wage; less than half of them have any employer-based health insurance coverage, and even fewer have pension plans. In all these social and economic characteristics, hospital NAs fare better (Crown et al., 1995).

As stated earlier in this report, because of the Medicare Prospective Payment System (PPS) and advances in medical technology, the acuity level of nursing home residents and the complexity of care provided are increasing. Changes in the characteristics of the residents, low staffing and low wages place added demands on the care givers, especially NAs.

The public hearings and site visits provided information from some nursing home personnel who worked at facilities with low NA staffing. In some nursing homes there is a clear need for more NAs to provide bedside care. Such situations were reported to have resulted in consequences such as failure to turn patients as required and decubitus ulcer formation; these may require hospitalization and other costly medical interventions. Furthermore, NAs are responsible for turning the patients in bed at regular intervals. If this is not done, contactures are likely to develop. NAs also feed patients and can be their primary source of companionship and psychological support. Thus, inadequate NA staffing leads to increased risk of medical complications and expense, intermittent discomfort from hunger and thirst, escalated need for even more nursing care, and sensory and psychological deprivation.

On the other hand, the committee also received testimony and saw nursing homes in which NA staffing was exemplary. One person who provided testimony to the committee felt strongly that the role of nurse assistants should be recreated to make it their primary mission to assist the resident rather than to assist other staff. At the facility with which this NA is affiliated, nurse assistants are called "resident assistants," and each is assigned to care for "families" of approximately five residents from the time the residents rise to the time they retire for the night. This witness strongly urged that the focus of NAs' work be shifted in this way and also urged that their titles be changed to "resident assistant'' to emphasize this new role. Innovative and successful programs decrease barriers between staff and patients, emphasize meeting residents' needs, increase respect for the work of the NAs, provide continuing education for NAs, experiment with flexible staffing patterns and tasks, and hold teams accountable for all aspects of residents' care. These programs seem to be associated with noteworthy resident, family, and staff satisfaction and with decreased staff turnover or use of sick leave. These remarkable institutions also find decreased medical problems that lead to resident hospitalizations, decreased incontinence, decreased use of restraints, and increased likelihood of residents walking or being able to make easy transfers.

Overall, on the basis of experience and information gathered from the testimonies received, the committee believes that the organization, use, and education of NA staff members make a substantial difference in the humane care, comfort, and health of nursing home residents and in the satisfaction and health of nursing staff.

In general, nurse assistants—who provide the largest portion of direct personal care to residents—receive little training for provision of care in a nursing facility. OBRA 87 requires 75 hours of training and testing for competency for nurse assistants within 4 months of employment and 12 hours of in-service training per year (sec. 1819(b)(5)). Federal regulations require that each state maintain a registry of NAs, but the exact nature of the training, certification, and requirements varies by states. Also, some states include NAs in their registry based on reciprocity with other states. With the increasing acuity of residents in nursing facilities and the complexity of care needed today, some argue that additional training tied to the clinical problems identified in nursing facilities is desirable. The committee heard testimony from certified nurse assistants working in nursing homes about the need for more clinical training and experience as part of their program leading to certification, as well as for the development of career ladders for NAs. There is also a management issue of provision of continuing on-the-job training and one-on-one guidance. Unfortunately, research is lacking on the effect of nurse assistant staffing and training on quality of care in nursing homes. As stated above, in public testimony and on site visits, however, the committee heard support for this relationship.

RECOMMENDATION 6-3: The committee recommends that the training for nurse assistants in nursing homes be structured and enriched by including training of the following types: appropriate clinical care of the aged and disabled; occupational health and safety measures; culturally sensitive care; and appropriate management of conflict.

RECOMMENDATION 6-4: The committee recommends that research efforts on staffing levels and skill mix specifically address the relationship of licensed practical nurses and nurse assistants to quality of care.

Management and Leadership

The changing focus of services and the increasingly complex nature of the care provided in nursing facilities place new demands for skills, judgment, supervision, and management of nursing services. Concern has been expressed by a number of nursing leaders about the training and educational preparation of RNs working in nursing facilities and especially of the directors of nursing (DON). Ballard (1995) indicates that the role of the DON or a nurse administrator ideally involves knowledge of nursing, management, organization theory, finance, marketing and planning, personnel administration, supervision, and government regulations. Most DONs in nursing facilities are not academically prepared for their positions (Bahr, 1991), having little or no specific education about the aging process, gerontological nursing principles, or managerial skills. In contrast to hospitals, where DONs only rarely have less than a bachelor's degree and often have graduate education, those in nursing homes are often graduates of associate degree and diploma programs in which leadership and management are not part of the basic preparation, and they rarely have advanced clinical training in gerontology. Again, this comparison is at the national aggregate level. Wide rural–urban variations can be found in the educational levels of RNs in managerial positions in hospitals.

Turnover among DONs in nursing facilities is high, amounting to more than 36 percent annually (AHCA, 1995), their salaries are low in comparison to hospitals, and they have limited opportunities for advancement. None of these factors is conducive to strong leadership. However, in view of the number of employees, budgets, and complexity of the care in nursing facilities today, strong leadership from the DON is a prerequisite for provision of high-quality, cost-effective care. Therefore:

RECOMMENDATION 6-5: The committee recommends that, in view of the increasing case-mix acuity of residents and the consequent complexity of the care provided, nursing facilities place greater weight on educational preparation in the employment of new directors of nursing.

In this regard, the committee is of the opinion that a bachelor's degree in nursing with special training in management and gerontology should be the preferred credential. In particular DONs need training in the management and administration of nursing facilities. The committee also urges that such facilities ensure a commitment to continuing education.

Staff Turnover

Nursing homes with higher NA-to-bed ratios and those that include nursing assistants as part of the care team, value their opinions, and acknowledge their important role in provision of quality care have lower turnover rates (Reagan, 1986; Wagnild and Manning, 1986; Willcocks et al., 1987; Wagnild, 1988; Birkenstock, 1991; Robertson et al., 1994; Mor, 1995). As discussed earlier in this chapter, information gathered from site visits, testimony, and small group meetings with DONs and others suggests that some facilities have reduced turnover by providing free on-site child care, health insurance, and other benefits. High RN and LPN turnover is associated with lower quality of care (Erickson, 1987; Wright, 1988; Munroe, 1990; Spector and Takada, 1991). More specifically, high turnover compromises the continuity of care and supervision of staff. Job turnover is also costly in terms of hiring, training, and facility productivity losses, but most important, high turnover rates adversely affect residents who do not cope well with frequent changes in staff (McDonald, 1994). Excessive turnover of these personnel, heavy use of part-time staff, and the use of floating or agency staff also compromise the quality of care (Erickson, 1987).

Permanent assignment of staff to residents results in more quality outcomes for residents and greater satisfaction and feelings of accountability for employees (Patchner and Patchner, 1993). Evaluation of a primary care model of delivery of nursing aide care (e.g., permanent aide assignment, a team approach, and enhanced communication) in nursing homes demonstrated increased quality-of-care indicators such as improved behavior, affect, and social activities among residents (Teresi et al., 1993).

National data on turnover indicate very high rates for all types of nursing personnel in nursing homes, especially nurse assistants (see Table 6.5). Moreover, staff turnover rates appear to have increased in recent years.

TABLE 6.5

Average Turnover Rates in Nursing Facilities by Staff Category, United States, 1990–1994.

Compensation

High rates of turnover in nursing homes are attributable to several causes. The low rate of compensation, compared to hospitals, has been a factor. In 1992, RNs' annual earnings in nursing homes were 14–17 percent below those in hospitals (Moses, 1994). On the one hand, RNs in nursing homes have, on the average, less educational preparation than those in hospitals. On the other hand, RNs in nursing facilities are much more likely to be employed in administrative positions than are those in hospitals (24 versus 3 percent in 1992) (Moses, 1994). Wages of nursing assistants are generally near the minimum wage and are comparable to levels offered by fast food chains and retail establishments. As with RNs, nursing assistants are paid appreciably less in nursing homes than in hospitals (Gold, 1995) (see Table 6.6), and they lag behind NAs in home health care agencies as well (Hospital and Healthcare Compensation Service, 1994).

TABLE 6.6

Nursing Facility Hourly Wages by Staff Category, United States, 1990–1993.

Many nursing facilities do not provide their employees with health benefits. Recently, AHCA (1994a) estimated that if mandatory national health insurance were adopted by Congress, nursing facility costs passed on to Medicaid would increase by $1 billion, and similar costs to Medicare would increase by $100 million. The 1994 average health insurance costs for nursing facilites were estimated to be 4 to 6 percent of the payroll. If all employees were provided health benefits, the health insurance costs would increase to almost 8 percent of the payroll (AHCA, 1994a).

Under a number of assumptions about the behavior of nursing facilities, higher levels of RN compensations result in reduced nursing home demand for RNs. Using data from 14,000 nursing facilities in 1987, Zinn (1993b) found that nursing facilities adjust staffing and care practices to local market conditions, as would be expected. In areas where RN wages were higher, nursing facilities employed more nonprofessional nursing staff. Thus, after controlling for resident characteristics, nursing facilities have economic incentives to hire fewer RNs in areas with high RN wages.

Clearly, the combination of low average wages and benefits contributes to high turnover and poor quality of care. During its site visits and in public testimony the committee heard many comments about the low level of wages and fringe benefits in nursing facilities, with the result that recruiting and retaining nurses are major problems for nursing homes. The committee is sympathetic to the need for increased compensation as a means of improving care. To achieve parity with other providers such as hospitals would increase the cost of care especially to Medicare and Medicaid. Higher compensation can possibly reduce the demand for RNs in nursing homes. It depends, however, on state Medicaid reimbursement methods and the internal resource allocation priorities that are established by nursing homes themselves. These quality and cost trade-offs must be considered in addressing this major problem in nursing homes.

Method and Level of Reimbursement

As discussed in Chapter 3, nursing homes derive most of their revenue from charges to private-pay patients and from Medicaid. Conceptually, both the level and the method of Medicaid reimbursement are determinants of nurse staffing levels. Traditionally, Medicaid is paid on a retrospective cost basis. Under this form of reimbursement, payment is made on the basis of costs incurred. This approach has been rapidly supplanted by other methods in which some or all of the rate is set prospectively (Swan et al., 1993a,b). Prospective-class (flat-rate) methods set prospective rates for groups of nursing homes within a state. Prospective facility-specific methods set rates by facility, generally using cost reports from earlier periods. Some states set rates prospectively but allow for retroactive adjustments (Swan et al., 1993a,b).

There is limited empirical evidence on the effect of level and method of Medicaid reimbursement on nurse staffing in nursing homes. Cohen and Spector (in press), using data from the 1987 National Medical Expenditure Survey, found that states with higher Medicaid reimbursement had more LPNs per 100 residents, adjusting for case-mix. However, a statistically significant effect was not obtained for RN staffing. They further found that Medicaid cost-based reimbursement led to substitution of RNs for LPNs. Presumably because payment is lower and there is excess demand for care on the part of Medicaid eligibles, facilities with high proportions of residents on Medicaid tend to have a lower quality of care as measured by process indicators (Nyman, 1985, 1989b; Gertler, 1989). Elderly persons who are potentially eligible for Medicaid have experienced access barriers to nursing home care in areas where a high proportion of potential nursing home residents are private (Ettner, 1993). One suggestion has been to tie the Medicaid reimbursement rate to the proportion of private patients in the home (Nyman, 1989b).

Case-Mix Reimbursement

Case-mix reimbursement attempts to tie payment to a facility's case-mix severity. Case-mix reimbursement systems were developed for Medicaid as a means of making closer linkages among resident needs, payments, and costs and as a way of removing access barriers for heavy-care Medicaid patients (Schlenker et al., 1985; Schlenker, 1991a,b). As noted above, 19 states were using case-mix systems in 1993 (Swan et al., 1994). The most commonly used case-mix measure has been functional status (using activities of daily living), although other disability scales have been used (Weissert and Musliner, 1992a,b). As mentioned earlier, one of the best known approaches has been the RUGS methodology developed by Fries and Cooney (1985), which has been updated into RUGS II and RUGS III versions (Fries et al., 1994). Resident characteristics are typically examined for the amount of personnel resources needed to provide care to residents, which can be determined in different ways such as staff time and cost studies (Weissert et al., 1983; Fries and Cooney, 1985; Arling et al., 1987; Fries et al., 1989, 1994). Once costs are determined, they are tied to resident characteristics (Weissert and Musliner, 1992a,b). As Fries and colleagues (1994) point out, the development of classification systems and resource use groups is primarily a technical process, but the development and assignment of reimbursement categories is primarily a political process.

Several studies have been conducted of case-mix (Weissert et al., 1983; Cameron, 1985; Fries and Cooney, 1985; Arling et al., 1987, 1989; Schneider et al., 1988; Fries et al., 1994). Weissert and Musliner (1992a,b) have summarized the results of the many studies of case-mix reimbursement. These studies reported that most states that have used case-mix reimbursement have improved access for some heavy-care residents (Ohio, Illinois, Maryland, and New York). On the other hand, there continued to be problems with access in some case-mix reimbursement states such as West Virginia (Holahan, 1984; Butler and Schlenker, 1988; Weissert and Musliner, 1992a,b). Access problems under case-mix, such as lengthy waiting lists for admissions, have occurred especially in areas where there is a low supply of beds (Nyman, 1988b), where there are Medicaid processing delays (Weissert and Cready, 1988), and where reimbursement rates are low. Access problems occurred for those with low-care needs and where community-based alternatives were not necessarily available (Butler and Schlenker, 1988; Feder and Scanlon, 1989).

Critical to the success of case-mix reimbursement is the adequacy of the case-mix measures themselves. The committee construes analysis of the underlying technical issues to be beyond the scope of its charge. (There is an extensive literature on this subject. See, for example, Fries and Cooney, 1985; Hu et al., 1986; Rohrer et al., 1989; Fries et al., 1994.) However, although attempting to base payment on severity is meritorious in principle, there may be problems in implementation. Classification errors may actually discourage delivery of quality therapeutic care, for example, if the system does not adequately account for comorbidities such as behavioral problems stemming from mental illness (Rohrer et al., 1989).

Case-mix reimbursement generally has not led to increases in nursing staff-to-resident ratios. In Maryland, there was no evidence that extra nursing home payments were used to add more staff (Feder and Scanlon, 1989). New York also did not increase staff even though resident case-mix increased (Butler and Schlenker, 1988). Although West Virginia had some evidence of poor quality (e.g., increased catheterization), nursing resources did increase in 1979–1981 (Holahan and Cohen, 1987; Weissert and Musliner, 1992a,b). In the San Diego experiment, where facilities were given financial incentives to take more heavy-care residents, there was no evidence that extra payments were spent on extra care (Meiners et al., 1985). Of the six state systems reviewed by Weissert and Musliner (1992a,b), only Illinois was rated as having improved quality (Holahan, 1984; Butler and Schlenker, 1988).

HCFA is undertaking a demonstration project to introduce Medicaid case-mix reimbursement in four states in 1994–1995. As Weissert and Musliner (1992a) have noted, it is not clear whether substantial new advances will be made in designing improved case-mix reimbursement systems in the demonstration project. An evaluation has been planned that will examine the outcomes of the demonstration on access, quality, and costs.

Incentives to Enhance Quality of Care

A conceptually attractive alternative to basing payment on inputs is to reward nursing homes based on aggregate outcomes achieved (Willemain, 1980; Kane et al., 1983). There have been several experiments with outcomes-based incentives. A social experiment was conducted in San Diego to test the effectiveness of monetary incentives in improving the health of nursing home residents and reducing Medicaid expenditures. With data from the San Diego project, Norton (1992) used the Markov model to represent the resulting health changes of nursing home residents. He found that offering incentives for improved outcomes had beneficial effects on both the quality and the cost of nursing home care. Furthermore, nursing homes admitted more persons with severe disabilities. The savings came not from more efficient use of nursing homes, but rather from savings from earlier hospital transfers. The experience in other locations has been mixed, however. Although the Illinois quality incentive program appears to have succeeded in increasing the quality of care, the validity of the outcome measures was not established (Geron, 1991). Connecticut's system was discontinued because the program's goals were not reached (Geron, 1991). Maryland's system of paying facilities to turn and position patients to prevent decubitus ulcers and to pay for improvement in ADLs for 2 months has been rated as effective (Weissert and Musliner, 1992a,b). Michigan's effort has not been evaluated (Lewin/ICF, 1991).

The committee finds the concept of reimbursing for improved outcomes intriguing, but recognizes that the implementation issues require further analysis. Also, using outcomes-based incentives may have some practical limits. In such cases, consideration should be given to linking reimbursement to process measures known to be associated with high quality of care.

RECOMMENDATION 6-6: The committee recommends that the Secretary of Health and Human Services fund additional research and demonstration projects on the use of financial and other incentives to improve quality of care and outcomes in nursing homes.