Do Not Hospitalize Directives in Long-term Care: Current Status and Practical Implications

In our society, death is medicalized, and quality of life is often equated with length of life. Because of this, it is more important than ever to keep end-of-life conversations at the forefront, especially when working in long-term care (LTC) facilities. Making time for end-of-life conversations with residents in LTC facilities promotes resident-centered decision-making by clarifying resident beliefs and values regarding end of life.¹ In LTC, such advance care planning conversations should be ongoing conversations with residents and their substitute decision-maker regarding what brings quality to their lives to ensure that future medical treatments align with their wishes.² Based on the literature, we know that most residents of Canadian LTC facilities do not want aggressive medical treatment.^3,4

Another related conversation that should be ongoing with residents in LTC facilities is clarifying their preferred place of treatment and preferred place of death. By and large, we know that residents of LTC facilities report better end-of-life comfort when they have not been hospitalized⁵ and that people prefer to die at home and not be transferred to a hospital.^3,6 This is likely due to unnecessary suffering and complications experienced by residents who are transferred to hospital settings for treatment where the transfer did not ultimately lead to an extension of life.^7,8 Although we know residents of LTC facilities generally do not want aggressive medical treatments, prefer to die at home, and are less comfortable in hospitals, findings from a systematic review of international research supported that one-third (33.2%) of all residents were still hospitalized during the last month of life, and 22.6% of residents still died in hospitals.⁹  

One strategy designed to mitigate unnecessary hospitalizations for residents of LTC facilities is the do not hospitalize (DNH) order. This is an advanced directive to which consent is given by a competent resident or their substitute decision-maker. It is important to establish that a DNH order does not mean “do not treat.” Instead, residents with a DNH order who have health concerns such as pneumonia, can be treated medically and remain in the LTC facility, but a resident requiring acute care treatment for a broken leg would be sent to a hospital. Other authors¹⁰ clarify that DNH directives are useful when hospital “transfers are unlikely to increase survival or improve patient quality of life.” Once a DNH directive is in place, its primary purpose is for the resident to receive required medical care in their LTC setting instead of going to the hospital. The advantages of DNH orders for residents of LTC facilities are well-supported in the literature to reduce adverse effects experienced by hospitalized residents such as gastrointestinal and respiratory infections, delirium, functional decline, pressure ulcers, and distress.⁷ Despite the advantages of DNH directives, only 21% of Canadian residents in LTC settings have one in place.¹¹

It is unclear why the prevalence of DNH directives in LTC facilities is so low. Results from one study show that substitute decision-makers are often confused about what a DNH means and do not understand the prognosis of the resident.¹⁰ Other researchers suggest that low engagement in DNH directives may be due to inconsistent processes and frequencies of DNH assessments between LTC homes and health care providers in Ontario, Canada.¹² Although the Long-Term Care Trust Act¹³ provides direction to LTC facilities that controls and governs LTC homes in their operation, it does not standardize the approach to DNH directives for practitioners. One way to promote person-centered decision-making for residents of LTC facilities might be to standardize the processes and procedures associated with how and when to engage residents and their substitute decision-makers in DNH conversations. To promote best practices in DNH conversations, it may be useful to profile the residents who are currently engaged in DNH conversations to identify residents who might be missed in the DNH process in LTC settings.

In Ontario, LTC facilities are regulated and funded by the Government of Ontario. In turn, government agencies determine admission criteria and eligibility while managing the up to 3-year waitlist. Each home is licensed to operate by the provincial government, and they must follow the highly regulated Long-Term Care Trust Act.¹³ Rigid and routine quality monitoring and inspection of all LTC homes in Ontario are managed by government agencies, using a minimum data set (MDS) resident assessment instrument. The purpose of the MDS assessment tool is to provide health care facilities and government agencies with data about gaps in care and the needs of residents in LTC facilities to plan and deliver better care.

According to the Ontario Long-Term Care Trust Act, MDS data is collected on each resident every 3 months, and it is stored by the Canadian Institute of Health Information. Routine staffing at each of Ontario’s LTC facilities includes at least 1 registered nurse and multiple nonregulated staff.¹³ Family physicians and nurse practitioners are also intermittently in attendance to address nursing concerns, conduct intake appointments, address health care needs, provide palliative care, and engage in nonurgent care, such as predental and preoperative assessments.¹⁴ 

There is limited research describing the resident population who do and do not have DNH directives. McGilton and colleagues¹⁵ explored resident characteristics associated with a documented DNH directive in LTC in Ontario. The researchers found residents who have DNH documentation are more likely to be women, live in rural areas, widowed, speak English or speak French, and have worsening cognitive impairment.¹⁵ While this is helpful insight, the researchers only assessed sociodemographic characteristics of the resident, cognitive impairment score, and their health stability, but there are many other characteristics that have the potential to influence the care of the resident and a DNH discussion. In particular, there has been no analysis of the potential relationship between DNH directives and residents’ diagnosis, physical functioning, communication, and mood or behavior pattern. More research is needed on the variables associated with DNH orders in Ontario’s LTC facilities. 

A better understanding of the characteristics of residents of LTC facilities who have and do not have DNH directives has the potential to lead to recommendations about shared LTC-wide DNH policies and procedures that would have a direct impact on resident care. This may help standardize best practice pathways for DNH directives to ensure that all substitute decision-makers understand their loved ones’ prognosis and the meaning of DNH directives so that residents are not missed in routine assessment of DNH. 

The research question guiding this study was: “What are the characteristics of Ontario residents of LTC facilities who have signed DNH orders?”

This was a population-based retrospective cohort study and we employed a retrospective cohort study design. We analyzed data retrospectively to examine the association between resident characteristics and DNH directives. The sample consisted of 101,315 residents in LTC homes between April 1, 2019, and March 31, 2020. We received permission for data usage from the Canadian Institute of Health Information, and the Research Ethics Board at Laurentian University approved the study. 

Data from the Resident Assessment Instrument (RAI) MDS assessment tool was used in this study. It was developed by the interRAI network, which includes researchers and health care practitioners from more than 35 countries. The interRAI network hosts a variety of standardized clinical assessment tools, which are edited and tailored to produce country-specific items. The Canadian interRAI tool—used in LTC facilities in Ontario—is MDS Version 2.0.¹⁶ The MDS assessment form includes demographic information about each resident, such as age and sex. It also includes variables related to the level of resident functioning for activities of daily living, such as whether the resident is able to walk, dress, eat, and use the bathroom by themselves. It also captures resident behaviors like crying, lashing out, or withdrawing. The MDS includes resident medical diagnoses, categories of medications within the previous 7 days, and treatments such as physiotherapy. MDS data collected about existing conditions relate to health care changes such as acute episodes, deterioration, and end-of-life stage. Finally, the MDS tool collects administrative variables including facility code, city, and advanced directives such as DNH.

MDS data from all LTC residents in Ontario from April 1, 2019, to March 31, 2020, were analyzed for the presence or absence of a signed DNH order as the dependent variable and a range of independent variables found in the MDS dataset. The dataset included 372,469 cases. Given that the MDS must be completed on each resident at least every 3 months, some residents represent multiple cases. To capture the assessment of each resident only once, we removed duplicates and kept only the most recent assessment. The final result included 101,315 cases. 

We conducted a descriptive analysis to determine the frequencies of age, sex, urban or rural setting, north or south location, and DNH status. We then conducted a Chi-square analysis of independent variables to determine which variables were significantly associated with the completion of a DNH directive. We used variables significant with DNH status and conducted a binomial logistic regression analysis to determine the characteristics of residents who are most likely to be associated with a DNH.  

A total of 101,315 residents were included in the analysis. The mean age of residents was 86 years. Of this total, 27.1% of the residents signed a DNH directive. See Table 1 for frequencies of variables in this study. 

Binomial regression analysis was conducted to determine the profile of characteristics of residents in LTC facilities in Ontario who have a DNH directive. Once a listwise deletion of cases with missing items was completed, the total number of cases in the regression analysis was 38,758. The shaded variables in Table 2 display the significant (P < .05) variables in the regression analysis comparing DNH orders to demographic and MDS variables. The findings are presented in the same categories as the MDS form.  

The findings in this study offer a snapshot of the resident characteristics that are most associated with a DNH order. Findings from this study showed that residents who were less likely to have a DNH order were younger, had a do not resuscitate order, lived in southern Ontario, had delirium, long-term memory deficits, deterioration in cognitive status, changes in hearing, exhibited repetitive verbalizations, unrealistic fears, unpleasant moods in the morning, sad facial expression, were resistant to care, changes in independent bowel care, took analgesia medication, and had an unstable medical condition requiring a change in the level of care, or end-stage disease.

By contrast, residents who were more likely to have a DNH order were those who exhibited repetitive anxiety, worsening dementia, exhibited socially inappropriate/disruptive behavior, were verbally abusive, depressive, had a history of falls, took diuretic medication, or experienced an acute episode in their health.

There are several limitations to this study. There is the possibility of bias that was addressed by capturing each resident’s assessment only once, using the most recent assessment. The large sample size reduces sampling error, offers high power and effect, and captures a full range of variation within the population.¹⁷ Although the data abstraction instrument, the RAI MDS 2.0, has high validity and reliability scores,¹⁸ we do not have contextual data, such as discussions that occurred before signing/not signing the DNH, which limits the interpretation of the findings. Another limitation is that the staff collecting and coding the MDS data are usually not responsible for direct resident care in LTC facilities. There is potential for wide variation and inaccuracies caused by incomplete knowledge of the residents by these staff. It is possible the MDS data of residents may not be accurate.

We must be cautious in generalizing the findings from this study because the MDS data set does not capture the complexity of health conditions experienced by each of the participants. While some of the findings are clinically intuitive, others are not. For example, a flare of an acute condition was significantly more associated with a DNH, while end-stage disease was not; although, it would seem that the opposite should be true in practice. Some of the findings suggest that DNH orders are more closely associated with residents who are behaviorally challenged, such as verbally abusive and socially inappropriate/disruptive behavior. It is possible that the residents displaying behaviors are in the advanced stages of dementia, which is itself associated with higher DNH orders.¹⁵

The variation in these findings demonstrates the need for additional contextual evidence before any conclusions can be drawn. However, the mix and range of significant variables seen in this study may be a result of inconsistent policies and procedures for DNH that are used across the LTC settings from which MDS data is collected. In various geographic locations in Canada, several LTC facilities have their own approach to DNH discussions. For example, one LTC facility set up a palliative care committee to review the care of each palliative resident at the end of life and the appropriateness of DNH orders. In other facilities, staff have informal consultations with the physician about DNH orders. Inconsistent and unsystematic DNH assessments in LTC could mean that many residents who would benefit from DNH may not be receiving the opportunity. Thus, these results support the need for the development of standardized policies and processes for DNH assessments in LTC, particularly to mitigate avoidable hospitalizations for LTC residents. Another practice consideration is to enable MDS data to capture contextual information related to DNH assessments and the flexibility to combine or link different variables in the dataset, such as responsive behaviors, advancing dementia, and DNH.

The novel development of standard and consistent procedures for securing DNH orders for residents in LTC facilities must be based on evidence, but the state of knowledge in this area is very weak. Additional research is required to better understand why residents and/or their substitute decision-makers consent to DNH orders and why/when healthcare providers engage residents in the conversation. Additional longitudinal research will be helpful to understand the impact of hospitalization and treatment on the quality of life of residents in LTC facilities who were eligible for the DNH directives. This may inform education strategies for healthcare providers, residents, and substitute decision-makers. 

DNH directives have the potential to improve the quality of life and quality of death for residents in LTC facilities and their loved ones. The purpose of this study was to gather a better understanding of the characteristics of residents in LTC facilities who have and do not have DNH directives. Findings from this study show diverse characteristics of residents who had DNH directives. Upon first examination, there were no evident patterns to these characteristics. Recommendations from this study are aimed at developing a standardized best practice pathway for DNH orders to ensure that residents are not missed in regular and routine assessment of DNH. 

Authors: Lori Rietze, RN, PhD¹ • Roberta Heale, PHC-NP, DNP, PhD¹ • Kirandeep Kaur, RN, MSN¹

Affiliations:
¹Laurentian University, Sudbury, Ontario, Canada

Disclosures:
The authors report no relevant financial relationships.

Address correspondence to:
Lori Rietze, RN, PhD
Laurentian University
935 Ramsey Lake Road
Sudbury, Ontario P3E 2C6, Canada
Email: lrietze@laurentian.ca