安寧緩和醫療對生命末期照護品質及醫療費用之相關性探討─以肺癌病人為例

Abstract

背景：為提升癌末照護品質及有效利用急性病房資源，衛生福利部逐步推動居家、住院安寧療護及安寧共同照護及社居安寧服務，然而鮮少研究透過縱貫資料分析安寧緩和醫療使用的軌跡類型及對生命末期照護品質及費用的影響。本研究以肺癌為例分析生命末期照護品質趨勢、安寧緩和醫療使用軌跡類型及影響因素，進而探討安寧緩和醫療對生命末期照護品質及醫療費用的影響。 研究方法：本研究為一回溯性世代研究，透過癌症登記資料庫、健保資料庫及死亡登記資料庫進行分析。研究對象為2004年至2012年申報至癌症登記資料庫之新診斷肺癌並於2012年底死亡個案(N=65,552)，根據每位個案在死前30週每週安寧療護使用情形透過群體軌跡模式辨識出安寧療護軌跡類型，以生命末期積極照護指標代表個案生命末期的照護品質，醫療費用包括死前一、三、六個月及納入安寧療護後的總醫療費用，共變項為病人人口學及疾病特性、照護醫師及醫院特性。以年齡年代世代模式、多元羅吉斯迴歸、傾向分數加權及配對方法、廣義線性混和模式進行假說驗證。 研究結果：肺癌生命末期安寧療護使用率逐年增加，積極照護指標逐年下降包括最後14天化學治療、最後一個月住院超過一次、住院超過14天、使用加護病房、氣管插管或呼吸器、及心肺復甦術；逐年增加指標包括最後14天標靶治療、最後一個月急診超過一次、及在醫院死亡。群體軌跡模式歸納出四種軌跡類型：延遲安寧療護組(85.7%個案，未使用安寧療護或在最後一週使用)，晚期安寧療護組(6.9%個案，死前6週開始使用且頻率快速升高)、中期安寧療護組(4.8%個案，死前13周使用且頻率緩慢升高)、及早期安寧療護組(2.6%個案，死前30週已使用且頻率緩慢升高)。女性、年長者、病齡長、疾病嚴重度高、居住地都市化程度高，於通過癌症診療品質認證、有安寧病房資源及高肺癌照護服務量醫師醫院就診患者有較高機率使用安寧療護。傾向分數加權後之廣義線性混和模式分析結果發現安寧療護組在最後14天化學治療、標靶治療、最後一個月急診超過1次、使用加護病房、插管呼吸器及心肺復甦術的相對風險低於延遲安寧療護組，但住院超過一次或超過14天及在醫院死亡風險高於延遲安寧療護組。安寧療護組在最後一、三、六個月、及納入安寧療護後總醫療費用皆顯著低於延遲安寧組，但並非越早納入安寧療護醫療費用降低越多。 結論：肺癌生命末期照護隨著安寧療護政策推動有逐漸增加的安寧療護使用率，同時減少積極照護使用及降低醫療費用。然而標靶治療、急診及在醫院死亡比例逐年增加，而安寧療護組在最後一個月住院及醫院死亡比例高於延遲安寧療護組，顯示生命末期標靶治療或其他新藥的使用規範，及居家或社區安寧療護制度仍需更完善的規劃，以提升肺癌患者安寧療護接受度及達到就地善終的目標。未來相關研究可透過群體軌跡模式作為客觀分類縱貫性資料次群體方法的參考。

Background: Hospice palliative medicine in Taiwan has been growing continuously since the National Health Insurance reimbursement scheme of hospice care was introduced to promote the quality of end-of-life (EOL) cancer care and to use the acute care resources efficiently. However, studies using longitudinal data to identify hospice palliative medicine trajectories and the impact on EOL care and medical expenditures were scant. This study aimed to explore the trend of EOL care quality among patients with lung cancer, to identify the trajectory patterns of hospice palliative medicine utilization and the factors related to these trajectories, and to evaluate the impact of hospice palliative medicine trajectories on EOL care quality and medical expenditures. Methods: A longitudinal retrospective cohort study was conducted by using nationwide database including Taiwan Cancer Registry, National Health Insurance Research Database and Taiwan Death Registry. Study population was 65,552 newly diagnosed lung cancer patients between 2004 and 2012 and died at the end of 2012. Hospice utilization of each patient at each week during 30 weeks before death was analyzed to identify the trajectory patterns by applying group-based trajectory modeling. Aggressive care indicators were assessed to represent the quality of EOL care. Medical expenditures were calculated at one, three, six months before death, and after enrollment in hospice programs. Information about patient demographics, disease characteristics, physician and hospital characteristics were also collected. Age-period-cohort models, multinomial logistic regression models, propensity score weighting and matching methods, and generalized linear mixed models were used to estimate associations. Results: Downward trends of aggressiveness EOL lung cancer care were observed for chemotherapy use, multiple hospitalizations, hospitalization exceeding 14 days, ICU admissions, intubation or mechanical ventilation use, and CPR use; upwards trends were observed for targeted therapy use, multiple emergency department visits, died in hospitals, and hospice palliative care use. Four distinct trajectory groups of hospice utilization in terminal lung cancer patients were identified: a late-referred group, a group with medium-referral and rapid rise in hospice utilization, a group with medium-referral and slow rise in hospice utilization, and an early-referral group. With late-referral as the reference group, those who were female, older age, longer disease duration, higher disease severity, living in higher urbanization area were more likely to use hospice care. The odds increased in patients whose treatment by physicians with a higher annual lung cancer case volume, and those whose treatment in hospitals that achieved the accreditation of cancer care quality, having inpatient hospice facilities, and with a higher annual lung cancer case volume. Multivariate analysis showed that medium- and early-referral patients were significantly less likely to undergo aggressive EOL care than late-referral patients including chemotherapy and targeted therapy use, multiple emergency room visits, ICU admissions, intubation, mechanical ventilation and CPR use. However, they were more likely to undergo multiple hospitalizations, hospitalization exceeding 14 days, and died in hospitals. Besides, medical expenditures of one, three, six months before death and after enrollment in hospice programs were significantly lower in medium- and early-referral groups than late-referral groups. However, not the earlier patient enrolled in hospice programs the more savings in medical expenditures. Conclusions: As the implementation of hospice reimbursement scheme, use of hospice palliative medicine increased, aggressive EOL care and medical expenditures decreased for patients with lung cancer. However, there is still room for developing the regulations of targeted therapy and new medication use at EOL, as well as home-based and community-based hospice palliative care system to increase the use of hospice palliative care and to reach the goal of passing on at home. Moreover, future researches could apply group-based trajectory modelling for longitudinal data classification.