Commentary: Chang CL, Tang GJ, Wu CP, Pu C, & Chen H-C. The influence of nurse practitioner staffing on intensive care unit mortality.
From time to time, our astute readers provide insightful and helpful feedback on published items. The previous article received a thorough read and critical review when it first appeared in our Online Now feature. The authors responded, and now the feedback and response are provided for our readers. Both names are withheld by request. All studies contain weaknesses, and we believe that recognizing this and sharing the following dialog is a great illustration of this fact. The take-home message for readers is that the ability to not just read but to critically read the research literature is an essential step in drawing accurate conclusions that could lead to practice changes and thus impact patient outcomes.
Each critique from the reader is numbered. The authors' response follows in italics.
1. Comparison of dissimilar groups
The study compared three hospital settings-one large academic medical center and two smaller community hospitals. The ICU bed size, MD staffing models, and NP staffing models differed, making comparisons difficult. Although the authors indicate that they adjusted the mortality risks among the three hospitals, the patients differed in many respects to question whether any comparisons can even be made. As outlined in Table 2, the patients from the three hospitals differed with respect to gender, age, severity of illness (as measured by APACHE II scores), hospital length of stay, and diagnosis.
Response: The stated purpose of the study was "to investigate the influences of NPs on ICU care quality and mortality by comparing ICUs from hospitals of different scales and with different care types". The authors took care in several sections of the manuscript to point out that there were a number of differences between the three care settings. The intent was to be a first-of-its-kind study in the country (Taiwan) to perform comparisons in facilities with clearly different staffing patterns.
2. Provision of detail about NP providers
The authors do not describe the NP staffing models. They do not indicate what the NP role was-did they manage patients independently, co-manage with the MD team, etc. They do not report the NP to patient ratios. They report MD and resident to patient ratios but not NP to patient ratios. The NP models of care differed in that Hospital A had none, Hospital B had two NPs, and Hospital C had one NP only on the day shift. You cannot make meaningful comparisons with so many differences in patient characteristics and NP models of care.
Response: Agree this could have been more explicit. The authors mention that NPs in Taiwan are used when physician staffing is inadequate and there is information in the article about use of both physicians and NPs to staff patients, with the implication that patients were staffed with either an NP or a physician (except hospital A where there were no NPs).
3. Number of ICU beds reported in the study
It is unclear if Hospital A had 120 ICU beds as stated on page 7 in the text or 30 beds as outlined in Table 1.
Response: The authors note at the bottom of page 2 which ICUs in Hospital A were included in the study.
4. Results of NP care in comparison to MDs
The results and discussion sections focus on mortality and staffing ratios for MDs. In the first paragraph of the discussion section, they cite staffing with higher MD to patient ratios and lower mortality risk, but this says nothing about the results of the NP models of care (especially as hospital A had no NPs).
Response: The authors acknowledge and include results from all three hospitals. Other differences between Hospital A and the others include staffing levels overall and provider/pt. ratio. The authors indicate that NP providers are staffing patients just like the MD providers are. Where there are more providers, there are better outcomes. The authors also include other clinical factors (like the use of EGDT and lactate levels) as reasons for differences rather than provider types.
5. Basing results on the data
The authors cite mortality among patients with sepsis and assert that because the literature identifies that NPs adhere to guidelines more, that sepsis care was improved-however, they did not measure anything related to sepsis care or cite any NP led sepsis initiatives. They also cite in the section on NPs have higher compliance on page 7 that Hospital B had the most NPs and the highest use of ScvO2 monitors "and possibly the best compliance with the sepsis treatment guidelines"-this is unfounded as they did not measure anything with respect to NP sepsis care.
Response: Agree this measure only addresses sepsis mortality. The authors state: "[horizontal ellipsis] and possibly the best compliance with the sepsis treatment guidelines; therefore, the mortality and SMRs of patients with septic shock in hospital B were the lowest." This is a stretch from the data. The authors also state, "This study used ScvO2, one of the indicators of early goal-directed therapy (EGDT), as a marker for compliance with the sepsis treatment guidelines and found that hospital B used the ScvO2 monitor more frequently than did the other two hospitals (Table 2) and better achieved the EGDT goals. Among the three hospitals, the SMR of patients with septic shock in hospital B was the lowest and was the best. Patients with septic shock were divided into three groups based on their APACHE II score, and the mortality was compared in each group. The mortality of hospital B in all three groups was lower than that of the other two hospitals" so there is some evidence to support the statement.
6. Conclusions compared with data
The conclusion paragraph on page 8 is unfounded as the authors indicate "in summary, NPs are not worse than resident physicians regarding clinical care mortality in ICU patients"-They have no data to substantiate this statement because they did not compare residents and NPs. In the next sentence, "When caring for patients with conditions that have clinical treatment guidelines, NPS have better compliance with the guidelines, resulting in better patient prognosis, even in small-scale hospitals," they did not measure this; they only cited the study by Landsperger. They cannot generalize that their study results in any way reflect this finding. In the ending conclusion section, paragraph 2, they cite, "The quality of care provided by NPs and resident physician did not significant differ". They should not claim this because they did not compare resident and NP care-they only compared mortality rates in three hospitals that varied tremendously with respect to patient and ICU characteristics.
Response: The authors did not intend to indicate that ICU patients in each facility have the same outcomes. In this study, only the outcomes of the patients with septic shock were compared. It is indeed insufficient to use ScvO2 as the indicator of compliance. However, future research is needed as mentioned in the last paragraph on page 8.
The indicators of quality of care in this case include mortality and SMR. Other researchers have also used SMR as one of the factors to evaluate the quality of care. The B hospital has the lowest SMR of septic shock patients. However, this study does not perform a large-scale cross-comparison and cannot demonstrate statistical significance. Therefore, the article states that there is no obvious difference in the outcomes.
7. Conclusions related to guidelines
In the sentence "Regarding diseases with treatment guidelines, such as the septic shock treatment guidelines, when NPs complied with the guidelines, they effectively reduced the mortality rate and the SMR, even in smaller hospitals," the investigators did not measure anything related to NP guideline compliance or care for sepsis and cannot assert that this is a study finding.
Response: See above. Agree this is not based on process data, the researchers' own outcomes data are cited, as is literature about compliance with other acute care guidelines. Conclusion does not appear to be fully supported by the data.
8. References
On page 2 column 1 the authors cite references from 2004, 2006, 2008 when there are current references in the literature related to sepsis mortality and incidence rates. They also cite the sepsis guidelines with a 2004 reference-the latest guidelines were published in 2017.
Response: There are 22 references, and eight of them are more than 10 years old. While this proportion is high, the study is based on data collected four years ago. Sepsis guidelines: on page seven, the authors discuss the initial sepsis guidelines from the Surviving Sepsis campaign, which was in 2004. This is referring to the inception of the guidelines and not the most recent ones. [black small square]