By late March 2020, New York City (NYC) was the epicenter of the COVID-19 pandemic in the United States. Early on, it was evident that nursing home (NH) residents with advanced life-limiting illnesses were extremely susceptible to this deadly disease. Like other NHs, Coler Rehabilitation and Nursing Care Center was confronted with an initial high infection fatality rate (IFR) reaching 24.4% in March 2020.
The vulnerability of the institution’s resident population was underscored by its predominantly male composition, and Blacks or African Americans comprised more than half of the known demographic makeup, both major COVID-19 mortality risk factors.1–4 Moreover, other key social determinants of health associated with poor COVID-19 outcomes were prevalent in the resident population of this large public NH before their admission here, which included poverty, homelessness, substandard housing, inadequate nutrition, and smoking.4 In addition, other major psychosocial factors prevalent among the inner city resident population were mental illness, substance and alcohol abuse, and significant behavioral problems, all of which adversely impacted adherence to treatment, as in some cases, residents refused to even wear face masks and, at times, to allow COVID-19 polymerase chain reaction (PCR) testing.
Although a compelling need to mitigate the COVID-19 impact was felt by Coler staff, there were no effective measures to counter its severe respiratory complications, as supportive care was deemed the best available therapeutic option.5 Nevertheless, the staff knew the residents were also at risk to develop other serious but treatable secondary complications, which could be detected by careful clinical surveillance during the early phase of COVID-19.
As geriatric principles for the prevention and management of secondary complications (eg, dehydration and bacterial pneumonia) in NH residents are well established, a proactive approach for early COVID-19 identification was adopted to avert or promptly treat complications before progression. As the standard care for typical NH residents in this country involved taking vital signs just monthly, a key element of the plan would require taking vital signs for all of Coler’s residents, not just those who already tested COVID-19 PCR positive (PCR+), at least once/shift, along with O2 saturation measurements via pulse oximetry. The rationale was to maintain a heightened level of suspicion for this illness to identify early subtle signs and symptoms, which otherwise could be overlooked. Criteria were set for a positive temperature screen at 99°F or 1°F above baseline temperature. A COVID-19 care bundle was then developed, similar to care bundles used widely throughout health care settings.6
The purpose of this approach was to reduce both the IFR and the new infection rate (IR) via early identification of cases, before the patients became seriously ill, thus limiting their opportunity to spread the disease.
Setting and Participants
This performance improvement (PI) initiative was conducted at Coler Rehabilitation and Nursing Care Center, an 815-bed public NH in NYC with full-time medical staff. Coler is the largest of 5 postacute/long-term care facilities in NYC Health + Hospitals, the largest public health care system in the United States, which serves as a safety net for the uninsured and underserved in NYC. Participants were all 204 residents who tested COVID-19 PCR+ for the first time over a 4-month period between March 1, 2020 and June 30, 2020. This PI initiative also included a 2-month follow-up period from July 1, 2020 to August 31, 2020, when an additional 4 residents tested newly PCR+.
In April 2020, this PI initiative was implemented, using the Plan-Do-Study-Act cycle, a repetitive 4-phase methodology commonly employed in health care.7,8 The goal was to determine the effectiveness of this newly developed COVID-19 PI bundle in reducing the IFR and the new IR among Coler NH residents.
Phases of Plan-Do-Study-Act Cycle
Figure 1 shows that in the Plan phase, a PI team comprising physicians, nurses, allied health professionals, and quality improvement professionals was assembled to develop the COVID-19 bundle. The initiative was carried out over a 3-month period from April to June 2020. A ≥33% reduction in the IFR from the March baseline along with a new IR <1% were considered to be successful. In the Do phase, the bundle was implemented, and a monthly log of newly tested COVID-19 PCR+ residents was compiled. As COVID-19 PCR testing was the standard used for confirmation of this illness, only PCR+ confirmed cases were included in this study. In the Study phase, data gathered from the Do phase were analyzed. Both monthly and biweekly IFRs and IRs were calculated and compared to the expected Plan phase outcomes to determine whether the initiative was progressing effectively. In the Act phase, improvements were identified and standardized into a protocol. Based on the data analysis and planning, adjustments were integrated into the bundle to improve its effectiveness. Thereafter, a 2-month follow-up from July to August was established to evaluate the effectiveness and sustainability of these adjustments over time.
In conducting this PI initiative, (1) no identifiers were used; (2) only established and acceptable standards of care were part of the bundle; and (3) the findings were limited to the facility and not generalizable. Thus, this PI initiative did not constitute research and did not require Institutional Review Board (IRB) review.9
COVID-19 Bundle Components
Primary bundle components, which focused on early identification of COVID-19 and its timely evaluation included: (1) taking of vital signs for all NH residents at least once/shift, along with O2 saturation measurements via pulse oximetry; (2) frequent follow-up visits (≥5 times/wk for those symptomatic) by the clinical staff, which was temporarily increased, along with daily medical staff huddles and weekly discussion among PI team leaders; (3) timely diagnostic testing, usually within a few hours or sooner when warranted, for symptomatic residents; (4) identification of residents suitable for palliative care with MOLST (medical orders for life-sustaining treatment) discussions directed toward maintaining quality of life and avoiding non-beneficial interventions that could potentially hasten their demise; and (5) ramped-up COVID-19 resident testing.
Secondary bundle components that involved treatments were as follows: (1) administration of supplemental O2 to symptomatic residents to normalize O2 saturations; (2) additional oral fluids for residents at risk for dehydration and intravenous fluids for those with dehydration; (3) antibiotic treatment for suspected secondary infections; and (4) hospitalization for seriously ill residents without preexisting do not hospitalize orders.
Monthly and biweekly IFRs were calculated by dividing the number of NH residents who initially tested PCR+ during each interval and who later died by the total number of residents who initially tested PCR+ during that interval. Biweekly and monthly new IRs were calculated by dividing the number of NH residents who newly tested positive during each interval by the total NH census during that interval. Control charts for IFRs and IRs were used to evaluate improvements over time. Periods of 15-day intervals were selected from January 31, 2020, when there was no evidence of the virus, to September 11, 2020. The PI initiative’s success was determined by control chart rules, which signaled the effectiveness of introduced changes. The 2-sample proportion test was used to determine differences between pre (baseline) and post bundle intervention IFR and new IR values. The Pearson correlation coefficient was used to determine whether a correlation existed between the trends of IFR and new IR over time. The presence of an association between the progressive IFR decline and the cumulative experience of the care teams in treating the 208 NH residents who tested COVID-19 PCR+ for the first time between March and July 2020 was determined using the χ2 test of independence. A P value <0.05 was considered statistically significant.
Figure 2 shows the COVID-19 IFR among NH residents before and after bundle implementation in April 2020. In the baseline phase (pre bundle implementation), COVID-19 was initially detected during the March 16, 2020 to March 30, 2020 interval. The IFR from this interval was 24.4%, which also was the average for the month of March. During the intervention phase (bundle implementation), an immediate uptick in the IFR to 31.3% was observed during the March 31, 2020 to April 14, 2020 interval, followed by a steady decline that continued until the May 30, 2020 to June 13, 2020 interval when the IFR was zero. In the postintervention phase (follow-up), the IFR remained at zero until September 11, 2020. The peak IFR, which occurred in mid-April within 2 weeks of implementation, was anticipated during the planning phase. The presence in the control chart of 8 consecutive points below the centerline, signaling a shift in the process, indicated that the bundle succeeded in reducing the IFR. The overall IFR average from April to June was 12.3%, a 49.6% reduction from the 24.4% March baseline (2-sample proportion test; P < 0.05).
Figure 3 shows the monthly IFR trend from March to August 2020, compared to the calculated regional average from the 4 neighboring states of New Jersey, Connecticut, Pennsylvania, and Massachusetts as of June 2020, obtained from the Kaiser Family Foundation. Although the 24.4% March baseline IFR was above the 21% regional average, it dipped below it by the end of April after bundle implementation. Thereafter, the IFR declined each month, reaching zero in June and remaining such until completion of the initiative.
Figure 4 shows the new COVID-19 IR among NH residents before and after bundle implementation in April 2020. From a baseline IR of 7.7% during the March 16, 2020 to March 30, 2020 interval, the IR initially decreased to 3.1% and then peaked at 12.3% during the April 15, 2020 to April 29, 2020 interval. Thereafter, it declined to <1% during the May 30, 2020 to June 13, 2020 interval, remaining such until completion of the follow-up phase at the end of August when the facility became COVID-19 free. The presence in the control chart of 4 out of 5 consecutive points below 1 sigma signaled that there was a change and that the bundle succeeded in reducing the new IR. The overall IR average from April to June was 4.1%, a 46.8% reduction from the 7.7% March baseline (2-sample proportion test; P < 0.001).
As shown in Figures 2 and 4, the trends of the IFR and new IR showed differences in direction in consecutive intervals from March 31, 2020 to April 14, 2020 and from April 15, 2020 to April 29, 2020. Thereafter, however, the IFR and the new IR declined simultaneously, both reaching zero by the end of the PI initiative. Overall, a moderate positive correlation was found between these 2 trends (Pearson correlation coefficient; P = 0.02).
Figure 5 shows that an association existed between the progressive IFR decline and the cumulative experience of the care teams in applying the bundle to treat the 208 NH residents who tested COVID-19 PCR+ for the first time from March to July 2020. The 208 residents were divided into 4 quartiles and 2 halves, based on the sequential order in which they tested PCR+ from earliest to latest. Figure 5A shows that the IFR dropped by 9.6% from the first to the second quartile. Thereafter, the IFR dropped by 5.8% and by 9.6% from the second to the third and from the third to the fourth quartiles, respectively. Thus, from the first to the fourth quartile, there was an overall IFR decline of 92.4% (χ2 test; P = 0.003). Figure 5B shows that the IFR dropped by 15.4% between the earlier half and the latter half, an overall IFR decline of 69.7% (χ2 test; P = 0.002).
The major finding of this initiative was the reduction of the IFR and new IR among NH residents in keeping with its objectives. By mid-May 2020, the bundle was succeeding as both COVID-19 rates fell simultaneously for the first time. It is of interest that several key components already contained in the Coler bundle were later recommended in a Journal of the American Medical Directors Association article that provided policy recommendations for managing COVID-19 in skilled nursing facilities.10
In June 2020, bundle adjustments were made that included: (1) COVID-19 PCR testing ramped up to once per week for all residents, even those who were asymptomatic and (2) creation of a special COVID-19 NH unit on a separate floor with dedicated staff assigned to just these residents. After these adjustments took effect, by mid-June the IFR fell from 4% to zero, where it remained until the completion of the initiative. The new IR fell from approximately 5% at the end of May to slightly below 1% by mid-June and finally dropped to zero by the end of August during the postintervention phase. The overall downward trends of the IFR and the new IR showed a positive correlation, indicating that the bundle’s implementation was effective in reducing both rates (data not shown).
It is of interest that both declines coincided with the testing of all staff, initially twice per week, beginning in mid-May as per New York State Department of Health requirements, and implementation of weekly testing of all Coler residents starting in June 2020. Furthermore, as shown in Figure 5, there was a positive association between the progressive IFR decline and the cumulative experience of the care teams in treating those with COVID-19, again pointing to the importance of early interventions.
Although this PI initiative achieved its objectives, it was not intended to determine the clinical basis for its success because this would constitute research. As this initiative was based on well-known geriatric principles, the authors would like to speculate further to provide possible clinical explanations for its success. It is likely that the early identification of COVID-19 was the key, enabling the care teams to avert or promptly treat complications before progression.
Residents in NHs, for instance, are more susceptible to dehydration because of multiple age-related factors that can be exacerbated by intercurrent infections. Thus, when not diagnosed timely or inadequately treated, dehydration in the elderly is associated with high mortality rates.11 A key reason is that dehydration raises the risk of urinary tract infections,12 which in turn increases the likelihood of uroseptic shock and death.13 Thus, the bundle was geared to identify residents at risk for dehydration with the goal of encouraging them to drink extra fluids, along with monitoring them carefully for subtle signs of it that otherwise might escape notice.
The same concern held true for NH-acquired pneumonia, the second most common infection in NH residents and a major cause of mortality, as prompt diagnosis and treatment are essential.14,15 The taking of vital signs and O2 saturations at least once per shift for all residents and frequent follow-up clinical visits for those with COVID-19 symptoms were intended to detect early indications of possible bacterial pneumonia so that proper treatments could be initiated promptly.
Pulse oximetry, a widely available precise and inexpensive technology, was a key in this initiative as it was performed every time vital signs were taken for residents. Whether pulse oximetry contributed to their survival by identifying early on the phenomenon of silent hypoxia, unique to this illness,16 enabling them to be administered supplement O2 timely is not known and should be the focus of future investigation.
Based on the increased mortality risks associated with the demographic makeup of the Coler resident population,1–4 an IFR average much higher than the regional average would have been expected; the opposite, however, was observed in this initiative. Thus, in view of well-founded concerns raised about disparities in the availability of quality health care adversely impacting African Americans, and based on the encouraging results of this initiative, further clinical research into this critical area is warranted.
Although no unanticipated consequences occurred after implementation, 1 issue encountered from the onset was the unavailability of comparative NH data, as the New York State Department of Health did not publicly release the total number of COVID-19-infected NH residents statewide or the number of them who later died from COVID-19 in hospitals. Thus, Coler’s IFR and new IR could not be compared to other NHs in the NYC area, which was the epicenter of the COVID-19 pandemic in this country during much of the spring of 2020. Instead, data from the Kaiser Family Foundation for the 4 bordering states of New Jersey, Connecticut, Pennsylvania, and Massachusetts were used for comparative purposes.17 The combined calculated IFR average from these states as of June 2020 was 21%, which was higher than the 12.3% average reported by the study institution for the period from April to June 2020. During this time period, the combined regional average did not vary greatly, whereas the Coler IFR steadily declined. It should, however, be noted that in some of these states, data included other types of long-term care facilities where residents typically are less debilitated and not as vulnerable to COVID-19, which would be expected to result in lower IFRs.
Although the bundle was successful in the study institution, limitations exist with regard to its applicability elsewhere, which would require further review to confirm its effectiveness. Also, when the pandemic first struck NYC, testing supplies were limited as public health advisories recommended that influenza testing should be first performed and ruled out before testing for COVID-19. The net effect likely hindered early identification of cases, contributing to a falsely elevated IFR, as asymptomatic and mildly symptomatic cases may not have been tested. Finally, the authors underscore that this was a PI study and not intended to determine the clinical basis for its success, as that would constitute research, thus requiring an IRB approval before it could be undertaken. The authors, however, do intend to seek IRB approval to later carry out such research.
Conclusions and Implications
This PI initiative succeeded in reducing the COVID-19 IFR from its peak in mid-April 2020 to zero in June, and it remained such thereafter. Furthermore, it lowered the new IR from its peak in April down to <1% between June and July. No new cases were reported in August upon completion of the initiative.
The implications of this study are promising, as COVID-19 fatalities were greatly reduced at the study institution, despite key social determinants of health associated with poor COVID-19 outcomes that were prevalent in our resident population. Thus, it is likely that the bundle or components of it could be successfully applied elsewhere to similar at-risk NH populations after appropriate assessment, which the authors encourage. Modifications tailored to the particular institution, however, may be necessary.
Although highly effective COVID-19 vaccines in the United States are now being distributed to NH residents, there likely will be a significant time lag until enough doses of vaccine are available for distribution to vulnerable populations in other parts of the world. The simplicity of this bundle makes it well-suited for prompt application in those places where vaccines will not be widely available soon. In addition, as in many of these locales, COVID-19 testing resources may be limited, and taking vital signs more frequently may help to identify potential COVID-19 cases much earlier and prioritize them for testing so that successful treatment can be administered in a timely manner.
As an in-depth analysis of the clinical factors contributing to the bundle’s success was not this initiative’s objective, we intend to perform a research study to investigate the role played by these factors, as well as to determine whether raising the standard of care for all, when circumstances warrant, can help overcome key social determinants of health associated with poor COVID-19 outcomes.
The authors acknowledge Robert K. Hughes, Chief Executive Officer at Coler Rehabilitation and Nursing Care Center, Khoi Luong, DO, Director of Post-Acute Care at NYC Health + Hospitals, Yves-Rose Pascal, RN, Coler Chief Nursing Officer, and Monserrate Nieves-Martinez, Coler Chief Quality Officer for their support and guidance. They offer special thanks to Paula E. Lester, MD, for her critical analysis and insight in helping to shape and refine this article.
Conflicts of Interest
The authors have no conflicts of interest to disclose.
This study, which was undertaken at Coler Rehabilitation and Nursing Care Center, did not receive any funding from agencies in the public, commercial, or not-for-profit sectors.
Drs Finger, Rao, Sansone, Hazel, and Ms Silvestri-Tan made a substantial contribution to the concept and design, acquisition of data, and analysis and interpretation of data; drafted the article, reviewed it critically for important intellectual content and then revised it; and agreed to submit and approved the version to be published.