Whole cohort summary statistics.

Background
A COVID-19 hospital guideline was implemented across all 18 acute hospitals in Wales in March 2020, promoting ward management of COVID pneumonitis and data collected across the first 3 Waves of the pandemic (Wave 1 March 1st 2020 to November 1st 2020, Wave 2 November 2st 2020 to February 21st 2021 and Wave 3 June 1st 2021 to December 14th 2021). The aim of this paper is to compare outcomes for patients by admission setting and type of ventilatory support given, with a particular focus on CPAP therapy.
Methods
This is a retrospective observational study of those aged over 18 admitted to hospital with community acquired COVID-19 between March 2020 and December 2021. The outcome of interest was in-hospital mortality. Univariate logistic regression models were used to compare crude outcomes across the waves. Multivariable logistic regression models were used to assess outcomes by different settings and treatments after adjusting for Wave, age, sex, co-morbidity and deprivation.
Results
Of the 7,803 records collected, 5,887 (75.4%) met the inclusion criteria. Analysis of those cases identified statistically significant outcome improvements across the waves for all patients combined (Waves 1 to 3: 31.5% to 18.8%, p<0.01), all ward patients (28.9% to 17.7%, p<0.01), and all ICU patients (44.3% to 32.2%, p = 0.03). Sub group analyses identified outcome improvements in ward patients without any oxygen therapy (Waves 1 to 3: 22.2% to 12.7%, p<0.01), with oxygen therapy only (34.0% to 12.9%, p<0.01) and with CPAP only (63.5% to 39.2%, p<0.01). The outcome improvements for ICU patients receiving CPAP only (35.7% to 24.6%, p = 0.31) or invasive ventilation (61.6% to 54.6%, p = 0.43) were not statistically significant though the numbers being admitted to ICU were small. The logistic regression models identified important age and comorbidity effects on outcomes. The multivariable model that took these into account suggested no statistically significantly greater risk of death for those receiving CPAP on the ward compared to those receiving CPAP in ICU (OR 0.89, 95% CI: 0.49 to 1.60).
Conclusions
There were successive reductions in mortality in inpatients over the three Waves reflecting new treatments and better management of complications. Mortality for those requiring CPAP was similar in respiratory wards and ICUs after adjusting for differences in their respective patient populations.

### 研究背景
2020年3月，威尔士地区全部18家急症医院均推行新冠肺炎医院诊疗指南，以规范新冠肺炎肺炎的病房管理，并收集了疫情前3轮流行周期的相关数据（第1轮：2020年3月1日至2020年11月1日；第2轮：2020年11月2日至2021年2月21日；第3轮：2021年6月1日至2021年12月14日）。本研究旨在对比不同收治场景与呼吸支持方式下患者的临床转归，重点关注持续气道正压通气（Continuous Positive Airway Pressure, CPAP）治疗。

### 研究方法
本研究为回顾性观察性研究，纳入2020年3月至2021年12月期间，年龄≥18岁、因社区获得性新冠肺炎住院的患者。本研究的主要结局指标为住院期间死亡率。采用单变量logistic回归模型比较各流行周期的粗死亡率；在校正流行周期、年龄、性别、合并症与社会剥夺程度后，采用多变量logistic回归模型评估不同收治场景与治疗方案对患者结局的影响。

### 研究结果
本次研究共收集7803份病例记录，其中5887份（75.4%）符合纳入标准。对符合纳入标准的病例进行分析显示，所有患者整体的住院死亡率随流行周期推进均出现统计学意义上的显著改善（第1至3轮：31.5%降至18.8%，p<0.01）；普通病房患者整体（28.9%降至17.7%，p<0.01）及重症监护病房（Intensive Care Unit, ICU）患者整体（44.3%降至32.2%，p=0.03）亦呈现相同趋势。
亚组分析结果显示，未接受氧疗的普通病房患者（22.2%降至12.7%，p<0.01）、仅接受氧疗的普通病房患者（34.0%降至12.9%，p<0.01）以及仅接受CPAP治疗的普通病房患者（63.5%降至39.2%，p<0.01）的住院死亡率均显著降低。而仅接受CPAP治疗的ICU患者（35.7%降至24.6%，p=0.31）与仅接受有创通气的ICU患者（61.6%降至54.6%，p=0.43）的死亡率下降未达到统计学显著性，不过此类ICU收治病例数较少。
Logistic回归分析显示，年龄与合并症对患者结局存在显著影响。在校正上述混杂因素后，多变量模型结果提示，病房CPAP治疗患者的死亡风险与ICU CPAP治疗患者无统计学差异（比值比OR=0.89，95%置信区间CI：0.49~1.60）。

### 研究结论
本研究纳入的3轮疫情住院患者的死亡率依次降低，反映出治疗方案的迭代升级与并发症管理水平的提升。在校正患者基线特征差异后，需接受CPAP治疗的患者，其在呼吸病房与ICU中的住院死亡率无显著差异。